{"id":29079,"date":"2023-08-18T06:23:21","date_gmt":"2023-08-18T09:23:21","guid":{"rendered":"https:\/\/mindthegraph.com\/blog\/construct-in-research-copy\/"},"modified":"2024-12-05T15:47:43","modified_gmt":"2024-12-05T18:47:43","slug":"hypothesis-testing","status":"publish","type":"post","link":"https:\/\/mindthegraph.com\/blog\/cs\/testovani-hypotez\/","title":{"rendered":"Testov\u00e1n\u00ed hypot\u00e9z: Principy a metody"},"content":{"rendered":"

Testov\u00e1n\u00ed hypot\u00e9z je z\u00e1kladn\u00edm n\u00e1strojem pou\u017e\u00edvan\u00fdm ve v\u011bdeck\u00e9m v\u00fdzkumu k potvrzen\u00ed nebo zam\u00edtnut\u00ed hypot\u00e9z o parametrech populace na z\u00e1klad\u011b v\u00fdb\u011brov\u00fdch dat. Poskytuje strukturovan\u00fd r\u00e1mec pro vyhodnocen\u00ed statistick\u00e9 v\u00fdznamnosti hypot\u00e9zy a vyvozen\u00ed z\u00e1v\u011br\u016f o skute\u010dn\u00e9 povaze populace. Testov\u00e1n\u00ed hypot\u00e9z se \u0161iroce pou\u017e\u00edv\u00e1 v oborech, jako jsou nap\u0159. biologie, psychologie, ekonomie a in\u017een\u00fdrstv\u00ed.<\/strong> ur\u010dovat \u00fa\u010dinnost nov\u00fdch l\u00e9\u010debn\u00fdch postup\u016f, zkoumat vztahy mezi prom\u011bnn\u00fdmi a p\u0159ij\u00edmat rozhodnut\u00ed zalo\u017een\u00e1 na datech. Navzdory sv\u00e9mu v\u00fdznamu v\u0161ak m\u016f\u017ee b\u00fdt testov\u00e1n\u00ed hypot\u00e9z n\u00e1ro\u010dn\u00e9 na pochopen\u00ed a spr\u00e1vnou aplikaci.<\/p>\n\n\n\n

V tomto \u010dl\u00e1nku se sezn\u00e1m\u00edme s testov\u00e1n\u00edm hypot\u00e9z, v\u010detn\u011b jeho \u00fa\u010delu, typ\u016f test\u016f, jednotliv\u00fdch krok\u016f, \u010dast\u00fdch chyb a osv\u011bd\u010den\u00fdch postup\u016f. A\u0165 u\u017e jste za\u010d\u00e1te\u010dn\u00edk, nebo zku\u0161en\u00fd v\u00fdzkumn\u00edk, tento \u010dl\u00e1nek v\u00e1m poslou\u017e\u00ed jako cenn\u00fd pr\u016fvodce zvl\u00e1dnut\u00edm testov\u00e1n\u00ed hypot\u00e9z ve va\u0161\u00ed pr\u00e1ci.<\/p>\n\n\n\n

\u00davod do testov\u00e1n\u00ed hypot\u00e9z<\/strong><\/h2>\n\n\n\n

Testov\u00e1n\u00ed hypot\u00e9z je statistick\u00fd n\u00e1stroj, kter\u00fd se b\u011b\u017en\u011b pou\u017e\u00edv\u00e1 ve v\u00fdzkumu ke zji\u0161t\u011bn\u00ed, zda existuje dostatek d\u016fkaz\u016f pro potvrzen\u00ed nebo zam\u00edtnut\u00ed hypot\u00e9zy. Zahrnuje formulaci hypot\u00e9zy o parametru populace, sb\u011br dat a jejich anal\u00fdzu s c\u00edlem ur\u010dit pravd\u011bpodobnost pravdivosti hypot\u00e9zy. Je d\u016fle\u017eitou sou\u010d\u00e1st\u00ed v\u011bdeck\u00e9 metody a pou\u017e\u00edv\u00e1 se v cel\u00e9 \u0159ad\u011b obor\u016f.<\/p>\n\n\n\n

Proces testov\u00e1n\u00ed hypot\u00e9z obvykle zahrnuje dv\u011b hypot\u00e9zy: nulovou hypot\u00e9zu a alternativn\u00ed hypot\u00e9zu. Nulov\u00e1 hypot\u00e9za je tvrzen\u00ed, \u017ee mezi dv\u011bma prom\u011bnn\u00fdmi neexistuje \u017e\u00e1dn\u00fd v\u00fdznamn\u00fd rozd\u00edl nebo mezi nimi nen\u00ed \u017e\u00e1dn\u00fd vztah, zat\u00edmco alternativn\u00ed hypot\u00e9za p\u0159edpokl\u00e1d\u00e1 existenci vztahu nebo rozd\u00edlu. V\u00fdzkumn\u00edci shroma\u017e\u010fuj\u00ed \u00fadaje a prov\u00e1d\u011bj\u00ed statistickou anal\u00fdzu, aby zjistili, zda lze nulovou hypot\u00e9zu zam\u00edtnout ve prosp\u011bch alternativn\u00ed hypot\u00e9zy.<\/p>\n\n\n\n

Testov\u00e1n\u00ed hypot\u00e9z se pou\u017e\u00edv\u00e1 k rozhodov\u00e1n\u00ed na z\u00e1klad\u011b \u00fadaj\u016f a je d\u016fle\u017eit\u00e9 pochopit z\u00e1kladn\u00ed p\u0159edpoklady a omezen\u00ed tohoto procesu. Je nezbytn\u00e9 zvolit vhodn\u00e9 statistick\u00e9 testy a velikosti vzork\u016f, aby bylo zaji\u0161t\u011bno, \u017ee v\u00fdsledky jsou p\u0159esn\u00e9 a spolehliv\u00e9, a m\u016f\u017ee b\u00fdt pro v\u00fdzkumn\u00e9 pracovn\u00edky mocn\u00fdm n\u00e1strojem k ov\u011b\u0159ov\u00e1n\u00ed jejich teori\u00ed a p\u0159ij\u00edm\u00e1n\u00ed rozhodnut\u00ed zalo\u017een\u00fdch na d\u016fkazech.<\/p>\n\n\n\n

\"\"<\/a><\/figure>\n\n\n\n

Typy test\u016f hypot\u00e9z<\/strong><\/h2>\n\n\n\n

Testov\u00e1n\u00ed hypot\u00e9z lze obecn\u011b rozd\u011blit do dvou kategori\u00ed: jednov\u00fdb\u011brov\u00e9 testy hypot\u00e9z a dvouv\u00fdb\u011brov\u00e9 testy hypot\u00e9z. Pod\u00edvejme se bl\u00ed\u017ee na ka\u017edou z t\u011bchto kategori\u00ed:<\/p>\n\n\n\n

Testy hypot\u00e9z na jednom vzorku<\/strong><\/h3>\n\n\n\n

P\u0159i testov\u00e1n\u00ed hypot\u00e9zy na jednom vzorku shroma\u017e\u010fuje v\u00fdzkumn\u00edk \u00fadaje z jedn\u00e9 populace a porovn\u00e1v\u00e1 je se zn\u00e1mou hodnotou nebo hypot\u00e9zou. Nulov\u00e1 hypot\u00e9za obvykle p\u0159edpokl\u00e1d\u00e1, \u017ee mezi pr\u016fm\u011bry populace a zn\u00e1mou hodnotou nebo hypot\u00e9zou nen\u00ed \u017e\u00e1dn\u00fd v\u00fdznamn\u00fd rozd\u00edl. V\u00fdzkumn\u00edk pak provede statistick\u00fd test, aby zjistil, zda je zji\u0161t\u011bn\u00fd rozd\u00edl statisticky v\u00fdznamn\u00fd. N\u011bkter\u00e9 p\u0159\u00edklady test\u016f jednov\u00fdb\u011brov\u00fdch hypot\u00e9z jsou n\u00e1sleduj\u00edc\u00ed:<\/p>\n\n\n\n

Jednov\u00fdb\u011brov\u00fd t-test:<\/strong> Tento test se pou\u017e\u00edv\u00e1 ke zji\u0161t\u011bn\u00ed, zda se v\u00fdb\u011brov\u00fd pr\u016fm\u011br v\u00fdznamn\u011b li\u0161\u00ed od p\u0159edpokl\u00e1dan\u00e9ho pr\u016fm\u011bru souboru.<\/p>\n\n\n\n

<\/div>\n\n\n\n
\"\"
Prost\u0159ednictv\u00edm statstest.b-cdn.net<\/a><\/em><\/figcaption><\/figure>\n\n\n\n
<\/div>\n\n\n\n

Jednov\u00fdb\u011brov\u00fd z-test:<\/strong> Tento test se pou\u017e\u00edv\u00e1 ke zji\u0161t\u011bn\u00ed, zda se v\u00fdb\u011brov\u00fd pr\u016fm\u011br v\u00fdznamn\u011b li\u0161\u00ed od p\u0159edpokl\u00e1dan\u00e9ho pr\u016fm\u011bru populace, je-li zn\u00e1ma sm\u011brodatn\u00e1 odchylka populace.<\/p>\n\n\n\n

<\/div>\n\n\n\n
\"\"
Prost\u0159ednictv\u00edm statstest.b-cdn.net<\/a><\/em><\/figcaption><\/figure>\n\n\n\n
<\/div>\n\n\n\n

Testy dvou v\u00fdb\u011brov\u00fdch hypot\u00e9z<\/strong><\/h3>\n\n\n\n

P\u0159i dvouv\u00fdb\u011brov\u00e9m testu hypot\u00e9zy shroma\u017e\u010fuje v\u00fdzkumn\u00edk \u00fadaje ze dvou r\u016fzn\u00fdch populac\u00ed a vz\u00e1jemn\u011b je porovn\u00e1v\u00e1. Nulov\u00e1 hypot\u00e9za obvykle p\u0159edpokl\u00e1d\u00e1, \u017ee mezi ob\u011bma populacemi nen\u00ed \u017e\u00e1dn\u00fd v\u00fdznamn\u00fd rozd\u00edl, a v\u00fdzkumn\u00edk provede statistick\u00fd test, aby zjistil, zda je pozorovan\u00fd rozd\u00edl statisticky v\u00fdznamn\u00fd. P\u0159\u00edklady test\u016f dvou v\u00fdb\u011brov\u00fdch hypot\u00e9z jsou n\u00e1sleduj\u00edc\u00ed:<\/p>\n\n\n\n

T-test nez\u00e1visl\u00fdch vzork\u016f:<\/strong> <\/em>Tento test se pou\u017e\u00edv\u00e1 k porovn\u00e1n\u00ed pr\u016fm\u011br\u016f dvou nez\u00e1visl\u00fdch vzork\u016f, aby se zjistilo, zda se od sebe v\u00fdznamn\u011b li\u0161\u00ed.<\/p>\n\n\n\n

<\/div>\n\n\n\n
\"\"
Prost\u0159ednictv\u00edm statstest.b-cdn.net<\/a><\/em><\/figcaption><\/figure>\n\n\n\n
<\/div>\n\n\n\n

P\u00e1rov\u00fd t-test: <\/strong>Tento test se pou\u017e\u00edv\u00e1 k porovn\u00e1n\u00ed pr\u016fm\u011br\u016f dvou p\u0159\u00edbuzn\u00fdch vzork\u016f, nap\u0159\u00edklad v\u00fdsledk\u016f p\u0159ed testem a po testu u stejn\u00e9 skupiny subjekt\u016f.<\/p>\n\n\n\n

Obr\u00e1zek: <\/strong>https:\/\/statstest.b-cdn.net\/wp-content\/uploads\/2020\/10\/Paired-Samples-T-Test.jpg<\/p>\n\n\n\n

Souhrnn\u011b lze \u0159\u00edci, \u017ee jednov\u00fdb\u011brov\u00e9 testy hypot\u00e9z se pou\u017e\u00edvaj\u00ed k testov\u00e1n\u00ed hypot\u00e9z o jedn\u00e9 populaci, zat\u00edmco dvouv\u00fdb\u011brov\u00e9 testy hypot\u00e9z se pou\u017e\u00edvaj\u00ed k porovn\u00e1v\u00e1n\u00ed dvou populac\u00ed. Vhodnost pou\u017eit\u00ed testu z\u00e1vis\u00ed na povaze dat a zkouman\u00e9 v\u00fdzkumn\u00e9 ot\u00e1zce.<\/p>\n\n\n\n

Kroky testov\u00e1n\u00ed hypot\u00e9z<\/strong><\/h2>\n\n\n\n

Testov\u00e1n\u00ed hypot\u00e9z zahrnuje \u0159adu krok\u016f, kter\u00e9 v\u00fdzkumn\u00edk\u016fm pom\u00e1haj\u00ed ur\u010dit, zda existuje dostatek d\u016fkaz\u016f pro potvrzen\u00ed nebo zam\u00edtnut\u00ed hypot\u00e9zy. Tyto kroky lze obecn\u011b rozd\u011blit do \u010dty\u0159 kategori\u00ed:<\/p>\n\n\n\n

Formulace hypot\u00e9zy<\/strong><\/h3>\n\n\n\n

Prvn\u00edm krokem p\u0159i testov\u00e1n\u00ed hypot\u00e9z je formulace nulov\u00e9 a alternativn\u00ed hypot\u00e9zy. Nulov\u00e1 hypot\u00e9za obvykle p\u0159edpokl\u00e1d\u00e1, \u017ee mezi dv\u011bma prom\u011bnn\u00fdmi neexistuje v\u00fdznamn\u00fd rozd\u00edl, zat\u00edmco alternativn\u00ed hypot\u00e9za p\u0159edpokl\u00e1d\u00e1 p\u0159\u00edtomnost vztahu nebo rozd\u00edlu. Je d\u016fle\u017eit\u00e9 formulovat jasn\u00e9 a testovateln\u00e9 hypot\u00e9zy je\u0161t\u011b p\u0159edt\u00edm, ne\u017e p\u0159istoup\u00edte ke sb\u011bru dat.<\/p>\n\n\n\n

Sb\u011br dat<\/strong><\/h3>\n\n\n\n

Druh\u00fdm krokem je shrom\u00e1\u017ed\u011bn\u00ed relevantn\u00edch \u00fadaj\u016f, kter\u00e9 lze pou\u017e\u00edt k ov\u011b\u0159en\u00ed hypot\u00e9z. Proces sb\u011bru dat by m\u011bl b\u00fdt pe\u010dliv\u011b navr\u017een tak, aby vzorek byl reprezentativn\u00ed pro z\u00e1jmovou populaci. Velikost vzorku by m\u011bla b\u00fdt dostate\u010dn\u011b velk\u00e1, aby bylo mo\u017en\u00e9 z\u00edskat statisticky platn\u00e9 v\u00fdsledky.<\/p>\n\n\n\n

Anal\u00fdza dat<\/strong><\/h3>\n\n\n\n

T\u0159et\u00edm krokem je anal\u00fdza dat pomoc\u00ed vhodn\u00fdch statistick\u00fdch test\u016f. Volba testu z\u00e1vis\u00ed na povaze dat a zkouman\u00e9 v\u00fdzkumn\u00e9 ot\u00e1zce. V\u00fdsledky statistick\u00e9 anal\u00fdzy poskytnou informace o tom, zda lze nulovou hypot\u00e9zu zam\u00edtnout ve prosp\u011bch alternativn\u00ed hypot\u00e9zy.<\/p>\n\n\n\n

Interpretace v\u00fdsledk\u016f<\/strong><\/h3>\n\n\n\n

Posledn\u00edm krokem je interpretace v\u00fdsledk\u016f statistick\u00e9 anal\u00fdzy. V\u00fdzkumn\u00edk mus\u00ed ur\u010dit, zda jsou v\u00fdsledky statisticky v\u00fdznamn\u00e9 a zda podporuj\u00ed nebo zam\u00edtaj\u00ed hypot\u00e9zu. V\u00fdzkumn\u00edk by m\u011bl tak\u00e9 zv\u00e1\u017eit omezen\u00ed studie a mo\u017en\u00e9 d\u016fsledky v\u00fdsledk\u016f.<\/p>\n\n\n\n

B\u011b\u017en\u00e9 chyby p\u0159i testov\u00e1n\u00ed hypot\u00e9z<\/strong><\/h2>\n\n\n\n

Testov\u00e1n\u00ed hypot\u00e9z je statistick\u00e1 metoda, kter\u00e1 se pou\u017e\u00edv\u00e1 ke zji\u0161t\u011bn\u00ed, zda existuje dostatek d\u016fkaz\u016f pro potvrzen\u00ed nebo zam\u00edtnut\u00ed ur\u010dit\u00e9 hypot\u00e9zy o parametru populace na z\u00e1klad\u011b vzorku dat. P\u0159i testov\u00e1n\u00ed hypot\u00e9z se mohou vyskytnout dva typy chyb:<\/p>\n\n\n\n

Chyba typu I: <\/strong>K tomu doch\u00e1z\u00ed, kdy\u017e v\u00fdzkumn\u00edk zam\u00edtne nulovou hypot\u00e9zu, p\u0159esto\u017ee je pravdiv\u00e1. Chyba typu I je tak\u00e9 zn\u00e1m\u00e1 jako fale\u0161n\u011b pozitivn\u00ed.<\/p>\n\n\n\n

Chyba typu II:<\/strong> <\/em>K tomu doch\u00e1z\u00ed, kdy\u017e v\u00fdzkumn\u00edk nezam\u00edtne nulovou hypot\u00e9zu, p\u0159esto\u017ee je nepravdiv\u00e1. Chyba typu II je tak\u00e9 zn\u00e1m\u00e1 jako fale\u0161n\u011b negativn\u00ed.<\/p>\n\n\n\n

Pro minimalizaci t\u011bchto chyb je d\u016fle\u017eit\u00e9 pe\u010dliv\u011b navrhnout a prov\u00e9st studii, zvolit vhodn\u00e9 statistick\u00e9 testy a spr\u00e1vn\u011b interpretovat v\u00fdsledky. V\u00fdzkumn\u00edci by si tak\u00e9 m\u011bli uv\u011bdomit omezen\u00ed sv\u00e9 studie a p\u0159i vyvozov\u00e1n\u00ed z\u00e1v\u011br\u016f zv\u00e1\u017eit mo\u017en\u00e9 zdroje chyb.<\/p>\n\n\n\n

Nulov\u00e9 a alternativn\u00ed hypot\u00e9zy<\/strong><\/h2>\n\n\n\n

P\u0159i testov\u00e1n\u00ed hypot\u00e9z existuj\u00ed dva typy hypot\u00e9z: nulov\u00e1 hypot\u00e9za a alternativn\u00ed hypot\u00e9za.<\/p>\n\n\n\n

Nulov\u00e1 hypot\u00e9za<\/strong><\/h3>\n\n\n\n

Nulov\u00e1 hypot\u00e9za (H0) je tvrzen\u00ed, kter\u00e9 p\u0159edpokl\u00e1d\u00e1, \u017ee mezi dv\u011bma prom\u011bnn\u00fdmi neexistuje \u017e\u00e1dn\u00fd v\u00fdznamn\u00fd rozd\u00edl nebo vztah. Je to v\u00fdchoz\u00ed hypot\u00e9za, kter\u00e1 se pova\u017euje za pravdivou, dokud nen\u00ed dostatek d\u016fkaz\u016f pro jej\u00ed zam\u00edtnut\u00ed. Nulov\u00e1 hypot\u00e9za se \u010dasto zapisuje jako tvrzen\u00ed o rovnosti, nap\u0159\u00edklad \"pr\u016fm\u011br skupiny A se rovn\u00e1 pr\u016fm\u011bru skupiny B\".<\/p>\n\n\n\n

Alternativn\u00ed hypot\u00e9za<\/strong><\/h3>\n\n\n\n

Alternativn\u00ed hypot\u00e9za (Ha) je tvrzen\u00ed, kter\u00e9 nazna\u010duje existenci v\u00fdznamn\u00e9ho rozd\u00edlu nebo vztahu mezi dv\u011bma prom\u011bnn\u00fdmi. Je to hypot\u00e9za, kterou m\u00e1 v\u00fdzkumn\u00edk z\u00e1jem ov\u011b\u0159it. Alternativn\u00ed hypot\u00e9za se \u010dasto zapisuje jako tvrzen\u00ed o nerovnosti, nap\u0159\u00edklad \"pr\u016fm\u011br skupiny A se nerovn\u00e1 pr\u016fm\u011bru skupiny B\".<\/p>\n\n\n\n

Nulov\u00e1 a alternativn\u00ed hypot\u00e9za se vz\u00e1jemn\u011b dopl\u0148uj\u00ed a vylu\u010duj\u00ed. Pokud je nulov\u00e1 hypot\u00e9za zam\u00edtnuta, je alternativn\u00ed hypot\u00e9za p\u0159ijata. Pokud nulovou hypot\u00e9zu nelze zam\u00edtnout, alternativn\u00ed hypot\u00e9za nen\u00ed potvrzena.<\/p>\n\n\n\n

Je d\u016fle\u017eit\u00e9 si uv\u011bdomit, \u017ee nulov\u00e1 hypot\u00e9za nemus\u00ed b\u00fdt nutn\u011b pravdiv\u00e1. Je to pouze tvrzen\u00ed, kter\u00e9 p\u0159edpokl\u00e1d\u00e1, \u017ee mezi zkouman\u00fdmi prom\u011bnn\u00fdmi neexistuje \u017e\u00e1dn\u00fd v\u00fdznamn\u00fd rozd\u00edl nebo vztah. \u00da\u010delem testov\u00e1n\u00ed hypot\u00e9z je zjistit, zda existuje dostatek d\u016fkaz\u016f pro zam\u00edtnut\u00ed nulov\u00e9 hypot\u00e9zy ve prosp\u011bch hypot\u00e9zy alternativn\u00ed.<\/p>\n\n\n\n

Hladina v\u00fdznamnosti a hodnota P<\/strong><\/h2>\n\n\n\n

P\u0159i testov\u00e1n\u00ed hypot\u00e9z je hladina v\u00fdznamnosti (alfa) pravd\u011bpodobnost\u00ed chyby typu I, tj. zam\u00edtnut\u00ed nulov\u00e9 hypot\u00e9zy, i kdy\u017e je ve skute\u010dnosti pravdiv\u00e1. Nej\u010dast\u011bji pou\u017e\u00edvan\u00e1 hladina v\u00fdznamnosti ve v\u011bdeck\u00e9m v\u00fdzkumu je 0,05, co\u017e znamen\u00e1, \u017ee existuje 5% pravd\u011bpodobnost chyby typu I.<\/p>\n\n\n\n

Hodnota p je statistick\u00e1 m\u00edra, kter\u00e1 ud\u00e1v\u00e1 pravd\u011bpodobnost z\u00edsk\u00e1n\u00ed pozorovan\u00fdch v\u00fdsledk\u016f nebo extr\u00e9mn\u011bj\u0161\u00edch v\u00fdsledk\u016f, pokud je nulov\u00e1 hypot\u00e9za pravdiv\u00e1. Je to m\u00edra s\u00edly d\u016fkazu proti nulov\u00e9 hypot\u00e9ze. Mal\u00e1 p-hodnota (obvykle men\u0161\u00ed ne\u017e zvolen\u00e1 hladina v\u00fdznamnosti 0,05) nazna\u010duje, \u017ee existuje siln\u00fd d\u016fkaz proti nulov\u00e9 hypot\u00e9ze, zat\u00edmco velk\u00e1 p-hodnota nazna\u010duje, \u017ee neexistuje dostatek d\u016fkaz\u016f pro zam\u00edtnut\u00ed nulov\u00e9 hypot\u00e9zy.<\/p>\n\n\n\n

Pokud je p-hodnota men\u0161\u00ed ne\u017e hladina v\u00fdznamnosti (p alfa), pak nulov\u00e1 hypot\u00e9za nen\u00ed zam\u00edtnuta a alternativn\u00ed hypot\u00e9za nen\u00ed potvrzena.<\/p>\n\n\n\n

Pokud chcete snadno pochopiteln\u00e9 shrnut\u00ed hladiny v\u00fdznamnosti, najdete ho v tomto \u010dl\u00e1nku: Srozumiteln\u00e9 shrnut\u00ed \u00farovn\u011b v\u00fdznamnosti<\/a>.<\/p>\n\n\n\n

Je d\u016fle\u017eit\u00e9 poznamenat, \u017ee statistick\u00e1 v\u00fdznamnost nemus\u00ed nutn\u011b znamenat praktick\u00fd v\u00fdznam nebo d\u016fle\u017eitost. Mal\u00fd rozd\u00edl nebo vztah mezi prom\u011bnn\u00fdmi m\u016f\u017ee b\u00fdt statisticky v\u00fdznamn\u00fd, ale nemus\u00ed m\u00edt praktick\u00fd v\u00fdznam. Statistick\u00e1 v\u00fdznamnost nav\u00edc z\u00e1vis\u00ed mimo jin\u00e9 na velikosti vzorku a velikosti \u00fa\u010dinku a m\u011bla by b\u00fdt interpretov\u00e1na v kontextu pl\u00e1nu studie a v\u00fdzkumn\u00e9 ot\u00e1zky.<\/p>\n\n\n\n

Anal\u00fdza s\u00edly pro testov\u00e1n\u00ed hypot\u00e9z<\/strong><\/h2>\n\n\n\n

Anal\u00fdza s\u00edly je statistick\u00e1 metoda pou\u017e\u00edvan\u00e1 p\u0159i testov\u00e1n\u00ed hypot\u00e9z ke stanoven\u00ed velikosti vzorku pot\u0159ebn\u00e9 k detekci ur\u010dit\u00e9 velikosti \u00fa\u010dinku s ur\u010ditou \u00farovn\u00ed spolehlivosti. S\u00edla statistick\u00e9ho testu je pravd\u011bpodobnost spr\u00e1vn\u00e9ho zam\u00edtnut\u00ed nulov\u00e9 hypot\u00e9zy, pokud je nepravdiv\u00e1, nebo pravd\u011bpodobnost, \u017ee se vyhneme chyb\u011b typu II.<\/p>\n\n\n\n

Anal\u00fdza s\u00edly je d\u016fle\u017eit\u00e1, proto\u017ee pom\u00e1h\u00e1 v\u00fdzkumn\u00edk\u016fm ur\u010dit vhodnou velikost vzorku pot\u0159ebnou k dosa\u017een\u00ed po\u017eadovan\u00e9 \u00farovn\u011b s\u00edly. Studie s n\u00edzkou silou m\u016f\u017ee selhat p\u0159i odhalen\u00ed skute\u010dn\u00e9ho \u00fa\u010dinku, co\u017e vede k chyb\u011b typu II, zat\u00edmco studie s vysokou silou s v\u011bt\u0161\u00ed pravd\u011bpodobnost\u00ed odhal\u00ed skute\u010dn\u00fd \u00fa\u010dinek, co\u017e vede k p\u0159esn\u011bj\u0161\u00edm a spolehliv\u011bj\u0161\u00edm v\u00fdsledk\u016fm.<\/p>\n\n\n\n

Aby bylo mo\u017en\u00e9 prov\u00e9st anal\u00fdzu s\u00edly, mus\u00ed v\u00fdzkumn\u00ed pracovn\u00edci ur\u010dit po\u017eadovanou \u00farove\u0148 s\u00edly, hladinu v\u00fdznamnosti, velikost \u00fa\u010dinku a velikost vzorku. Velikost \u00fa\u010dinku je m\u00edrou velikosti rozd\u00edlu nebo vztahu mezi zkouman\u00fdmi prom\u011bnn\u00fdmi a obvykle se odhaduje na z\u00e1klad\u011b p\u0159edchoz\u00edho v\u00fdzkumu nebo pilotn\u00edch studi\u00ed. Anal\u00fdza s\u00edly pak m\u016f\u017ee ur\u010dit pot\u0159ebnou velikost vzorku, kter\u00e1 je nutn\u00e1 k dosa\u017een\u00ed po\u017eadovan\u00e9 \u00farovn\u011b s\u00edly.<\/p>\n\n\n\n

Anal\u00fdzu s\u00edly lze tak\u00e9 pou\u017e\u00edt zp\u011btn\u011b k ur\u010den\u00ed s\u00edly dokon\u010den\u00e9 studie na z\u00e1klad\u011b velikosti vzorku, velikosti \u00fa\u010dinku a hladiny v\u00fdznamnosti. To m\u016f\u017ee v\u00fdzkumn\u00fdm pracovn\u00edk\u016fm pomoci vyhodnotit s\u00edlu jejich z\u00e1v\u011br\u016f a ur\u010dit, zda je t\u0159eba prov\u00e9st dal\u0161\u00ed v\u00fdzkum.<\/p>\n\n\n\n

Celkov\u011b je anal\u00fdza s\u00edly d\u016fle\u017eit\u00fdm n\u00e1strojem p\u0159i testov\u00e1n\u00ed hypot\u00e9z, proto\u017ee pom\u00e1h\u00e1 v\u00fdzkumn\u00fdm pracovn\u00edk\u016fm navrhovat studie, kter\u00e9 jsou dostate\u010dn\u011b siln\u00e9, aby odhalily skute\u010dn\u00e9 \u00fa\u010dinky a vyhnuly se chyb\u00e1m typu II.<\/p>\n\n\n\n

Bayesovsk\u00e9 testov\u00e1n\u00ed hypot\u00e9z<\/strong><\/h2>\n\n\n\n

Bayesovsk\u00e9 testov\u00e1n\u00ed hypot\u00e9z je statistick\u00e1 metoda, kter\u00e1 umo\u017e\u0148uje v\u00fdzkumn\u00edk\u016fm vyhodnotit d\u016fkazy pro a proti konkuren\u010dn\u00edm hypot\u00e9z\u00e1m na z\u00e1klad\u011b pravd\u011bpodobnosti pozorovan\u00fdch dat p\u0159i ka\u017ed\u00e9 hypot\u00e9ze a tak\u00e9 na z\u00e1klad\u011b p\u0159edb\u011b\u017en\u00e9 pravd\u011bpodobnosti ka\u017ed\u00e9 hypot\u00e9zy. Na rozd\u00edl od klasick\u00e9ho testov\u00e1n\u00ed hypot\u00e9z, kter\u00e9 se zam\u011b\u0159uje na zam\u00edt\u00e1n\u00ed nulov\u00fdch hypot\u00e9z na z\u00e1klad\u011b p-hodnot, poskytuje bayesovsk\u00e9 testov\u00e1n\u00ed hypot\u00e9z diferencovan\u011bj\u0161\u00ed a informativn\u011bj\u0161\u00ed p\u0159\u00edstup k testov\u00e1n\u00ed hypot\u00e9z, nebo\u0165 umo\u017e\u0148uje v\u00fdzkumn\u00edk\u016fm kvantifikovat s\u00edlu d\u016fkaz\u016f pro a proti ka\u017ed\u00e9 hypot\u00e9ze.<\/p>\n\n\n\n

P\u0159i bayesovsk\u00e9m testov\u00e1n\u00ed hypot\u00e9z za\u010d\u00ednaj\u00ed v\u00fdzkumn\u00edci s p\u0159edb\u011b\u017en\u00fdm rozd\u011blen\u00edm pravd\u011bpodobnosti pro ka\u017edou hypot\u00e9zu, kter\u00e9 vych\u00e1z\u00ed z existuj\u00edc\u00edch znalost\u00ed nebo p\u0159esv\u011bd\u010den\u00ed. Pot\u00e9 aktualizuj\u00ed p\u0159edb\u011b\u017en\u00e9 rozd\u011blen\u00ed pravd\u011bpodobnosti na z\u00e1klad\u011b pravd\u011bpodobnosti pozorovan\u00fdch dat pro ka\u017edou hypot\u00e9zu pomoc\u00ed Bayesovy v\u011bty. V\u00fdsledn\u00e9 posteriorn\u00ed rozd\u011blen\u00ed pravd\u011bpodobnosti p\u0159edstavuje pravd\u011bpodobnost ka\u017ed\u00e9 hypot\u00e9zy vzhledem k pozorovan\u00fdm \u00fadaj\u016fm.<\/p>\n\n\n\n

S\u00edlu d\u016fkaz\u016f pro jednu hypot\u00e9zu ve srovn\u00e1n\u00ed s jinou lze kvantifikovat v\u00fdpo\u010dtem Bayesova faktoru, co\u017e je pom\u011br pravd\u011bpodobnosti pozorovan\u00fdch dat p\u0159i jedn\u00e9 hypot\u00e9ze a druh\u00e9, v\u00e1\u017een\u00fd jejich p\u0159edchoz\u00edmi pravd\u011bpodobnostmi. Bayes\u016fv faktor v\u011bt\u0161\u00ed ne\u017e 1 znamen\u00e1 d\u016fkaz ve prosp\u011bch jedn\u00e9 hypot\u00e9zy, zat\u00edmco Bayes\u016fv faktor men\u0161\u00ed ne\u017e 1 znamen\u00e1 d\u016fkaz ve prosp\u011bch druh\u00e9 hypot\u00e9zy.<\/p>\n\n\n\n

Bayesovsk\u00e9 testov\u00e1n\u00ed hypot\u00e9z m\u00e1 oproti klasick\u00e9mu testov\u00e1n\u00ed hypot\u00e9z n\u011bkolik v\u00fdhod. Zaprv\u00e9 umo\u017e\u0148uje v\u00fdzkumn\u00edk\u016fm aktualizovat sv\u00e1 p\u0159edchoz\u00ed p\u0159esv\u011bd\u010den\u00ed na z\u00e1klad\u011b pozorovan\u00fdch dat, co\u017e m\u016f\u017ee v\u00e9st k p\u0159esn\u011bj\u0161\u00edm a spolehliv\u011bj\u0161\u00edm z\u00e1v\u011br\u016fm. Za druh\u00e9 poskytuje informativn\u011bj\u0161\u00ed m\u00edru d\u016fkaz\u016f ne\u017e p-hodnoty, kter\u00e9 pouze ud\u00e1vaj\u00ed, zda jsou pozorovan\u00e1 data statisticky v\u00fdznamn\u00e1 na p\u0159edem stanoven\u00e9 hladin\u011b. V neposledn\u00ed \u0159ad\u011b dok\u00e1\u017ee pojmout slo\u017eit\u00e9 modely s v\u00edce parametry a hypot\u00e9zami, kter\u00e9 mohou b\u00fdt obt\u00ed\u017en\u011b analyzovateln\u00e9 pomoc\u00ed klasick\u00fdch metod.<\/p>\n\n\n\n

Celkov\u011b lze \u0159\u00edci, \u017ee bayesovsk\u00e9 testov\u00e1n\u00ed hypot\u00e9z je v\u00fdkonn\u00e1 a flexibiln\u00ed statistick\u00e1 metoda, kter\u00e1 m\u016f\u017ee v\u00fdzkumn\u00fdm pracovn\u00edk\u016fm pomoci \u010dinit informovan\u011bj\u0161\u00ed rozhodnut\u00ed a vyvozovat p\u0159esn\u011bj\u0161\u00ed z\u00e1v\u011bry z jejich dat.<\/p>\n\n\n\n

Vytv\u00e1\u0159ejte v\u011bdecky p\u0159esn\u00e9 infografiky b\u011bhem n\u011bkolika minut<\/strong><\/h2>\n\n\n\n

Mind the Graph<\/a> je v\u00fdkonn\u00fd n\u00e1stroj, kter\u00fd pom\u00e1h\u00e1 v\u011bdc\u016fm snadno vytv\u00e1\u0159et v\u011bdecky p\u0159esn\u00e9 infografiky. D\u00edky intuitivn\u00edmu rozhran\u00ed, p\u0159izp\u016fsobiteln\u00fdm \u0161ablon\u00e1m a rozs\u00e1hl\u00e9 knihovn\u011b v\u011bdeck\u00fdch ilustrac\u00ed a ikon umo\u017e\u0148uje Mind the Graph v\u011bdc\u016fm snadno vytv\u00e1\u0159et profesion\u00e1ln\u011b vypadaj\u00edc\u00ed grafiku, kter\u00e1 \u00fa\u010dinn\u011b zprost\u0159edkuje jejich v\u00fdsledky \u0161irok\u00e9mu publiku.<\/p>\n\n\n\n

<\/div>\n\n\n
\n
\"\"<\/a><\/figure><\/div>\n\n\n
<\/div>","protected":false},"excerpt":{"rendered":"

P\u0159e\u010dt\u011bte si informace o testov\u00e1n\u00ed hypot\u00e9z. Typy test\u016f, \u010dast\u00e9 chyby, osv\u011bd\u010den\u00e9 postupy a dal\u0161\u00ed informace. Ide\u00e1ln\u00ed pro v\u0161echny v\u00fdzkumn\u00edky.<\/p>","protected":false},"author":35,"featured_media":29081,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":[],"categories":[978,974,961],"tags":[],"yoast_head":"\nHypothesis Testing: Principles and Methods<\/title>\n<meta name=\"description\" content=\"Learn about hypothesis testing. The types of tests, common errors, best practices, and more. Perfect for all researchers.\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/mindthegraph.com\/blog\/cs\/testovani-hypotez\/\" \/>\n<meta property=\"og:locale\" content=\"cs_CZ\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Hypothesis Testing: Principles and Methods\" \/>\n<meta property=\"og:description\" content=\"Learn about hypothesis testing. The types of tests, common errors, best practices, and more. Perfect for all researchers.\" \/>\n<meta property=\"og:url\" content=\"https:\/\/mindthegraph.com\/blog\/cs\/testovani-hypotez\/\" \/>\n<meta property=\"og:site_name\" content=\"Mind the Graph Blog\" \/>\n<meta property=\"article:published_time\" content=\"2023-08-18T09:23:21+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2024-12-05T18:47:43+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/mindthegraph.com\/blog\/wp-content\/uploads\/2023\/08\/hypothesis-testing-blog.jpg\" \/>\n\t<meta property=\"og:image:width\" content=\"1123\" \/>\n\t<meta property=\"og:image:height\" content=\"612\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/jpeg\" \/>\n<meta name=\"author\" content=\"Ang\u00e9lica Salom\u00e3o\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:title\" content=\"Hypothesis Testing: Principles and Methods\" \/>\n<meta name=\"twitter:description\" content=\"Learn about hypothesis testing. The types of tests, common errors, best practices, and more. Perfect for all researchers.\" \/>\n<meta name=\"twitter:image\" content=\"https:\/\/mindthegraph.com\/blog\/wp-content\/uploads\/2023\/08\/hypothesis-testing-blog.jpg\" \/>\n<meta name=\"twitter:label1\" content=\"Written by\" \/>\n\t<meta name=\"twitter:data1\" content=\"Ang\u00e9lica Salom\u00e3o\" \/>\n\t<meta name=\"twitter:label2\" content=\"Est. reading time\" \/>\n\t<meta name=\"twitter:data2\" content=\"11 minutes\" \/>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"Hypothesis Testing: Principles and Methods","description":"Learn about hypothesis testing. The types of tests, common errors, best practices, and more. Perfect for all researchers.","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/mindthegraph.com\/blog\/cs\/testovani-hypotez\/","og_locale":"cs_CZ","og_type":"article","og_title":"Hypothesis Testing: Principles and Methods","og_description":"Learn about hypothesis testing. The types of tests, common errors, best practices, and more. Perfect for all researchers.","og_url":"https:\/\/mindthegraph.com\/blog\/cs\/testovani-hypotez\/","og_site_name":"Mind the Graph Blog","article_published_time":"2023-08-18T09:23:21+00:00","article_modified_time":"2024-12-05T18:47:43+00:00","og_image":[{"width":1123,"height":612,"url":"https:\/\/mindthegraph.com\/blog\/wp-content\/uploads\/2023\/08\/hypothesis-testing-blog.jpg","type":"image\/jpeg"}],"author":"Ang\u00e9lica Salom\u00e3o","twitter_card":"summary_large_image","twitter_title":"Hypothesis Testing: Principles and Methods","twitter_description":"Learn about hypothesis testing. The types of tests, common errors, best practices, and more. Perfect for all researchers.","twitter_image":"https:\/\/mindthegraph.com\/blog\/wp-content\/uploads\/2023\/08\/hypothesis-testing-blog.jpg","twitter_misc":{"Written by":"Ang\u00e9lica Salom\u00e3o","Est. reading time":"11 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/mindthegraph.com\/blog\/hypothesis-testing\/","url":"https:\/\/mindthegraph.com\/blog\/hypothesis-testing\/","name":"Hypothesis Testing: Principles and Methods","isPartOf":{"@id":"https:\/\/mindthegraph.com\/blog\/#website"},"datePublished":"2023-08-18T09:23:21+00:00","dateModified":"2024-12-05T18:47:43+00:00","author":{"@id":"https:\/\/mindthegraph.com\/blog\/#\/schema\/person\/542e3620319366708346388407c01c0a"},"description":"Learn about hypothesis testing. The types of tests, common errors, best practices, and more. Perfect for all researchers.","breadcrumb":{"@id":"https:\/\/mindthegraph.com\/blog\/hypothesis-testing\/#breadcrumb"},"inLanguage":"cs-CZ","potentialAction":[{"@type":"ReadAction","target":["https:\/\/mindthegraph.com\/blog\/hypothesis-testing\/"]}]},{"@type":"BreadcrumbList","@id":"https:\/\/mindthegraph.com\/blog\/hypothesis-testing\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/mindthegraph.com\/blog\/"},{"@type":"ListItem","position":2,"name":"Hypothesis Testing: Principles and Methods"}]},{"@type":"WebSite","@id":"https:\/\/mindthegraph.com\/blog\/#website","url":"https:\/\/mindthegraph.com\/blog\/","name":"Mind the Graph Blog","description":"Your science can be beautiful!","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/mindthegraph.com\/blog\/?s={search_term_string}"},"query-input":"required name=search_term_string"}],"inLanguage":"cs-CZ"},{"@type":"Person","@id":"https:\/\/mindthegraph.com\/blog\/#\/schema\/person\/542e3620319366708346388407c01c0a","name":"Ang\u00e9lica Salom\u00e3o","image":{"@type":"ImageObject","inLanguage":"cs-CZ","@id":"https:\/\/mindthegraph.com\/blog\/#\/schema\/person\/image\/","url":"https:\/\/secure.gravatar.com\/avatar\/a59218eda57fb51e0d7aea836e593cd1?s=96&d=mm&r=g","contentUrl":"https:\/\/secure.gravatar.com\/avatar\/a59218eda57fb51e0d7aea836e593cd1?s=96&d=mm&r=g","caption":"Ang\u00e9lica Salom\u00e3o"},"url":"https:\/\/mindthegraph.com\/blog\/cs\/author\/angelica\/"}]}},"_links":{"self":[{"href":"https:\/\/mindthegraph.com\/blog\/cs\/wp-json\/wp\/v2\/posts\/29079"}],"collection":[{"href":"https:\/\/mindthegraph.com\/blog\/cs\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/mindthegraph.com\/blog\/cs\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/mindthegraph.com\/blog\/cs\/wp-json\/wp\/v2\/users\/35"}],"replies":[{"embeddable":true,"href":"https:\/\/mindthegraph.com\/blog\/cs\/wp-json\/wp\/v2\/comments?post=29079"}],"version-history":[{"count":8,"href":"https:\/\/mindthegraph.com\/blog\/cs\/wp-json\/wp\/v2\/posts\/29079\/revisions"}],"predecessor-version":[{"id":55768,"href":"https:\/\/mindthegraph.com\/blog\/cs\/wp-json\/wp\/v2\/posts\/29079\/revisions\/55768"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/mindthegraph.com\/blog\/cs\/wp-json\/wp\/v2\/media\/29081"}],"wp:attachment":[{"href":"https:\/\/mindthegraph.com\/blog\/cs\/wp-json\/wp\/v2\/media?parent=29079"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/mindthegraph.com\/blog\/cs\/wp-json\/wp\/v2\/categories?post=29079"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/mindthegraph.com\/blog\/cs\/wp-json\/wp\/v2\/tags?post=29079"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}