{"id":537,"date":"2018-07-21T10:45:17","date_gmt":"2018-07-21T01:45:17","guid":{"rendered":"https:\/\/best-biostatistics.com\/toukei-er\/entry\/how-to-determine-sample-size-in-t-test\/"},"modified":"2024-12-24T18:35:48","modified_gmt":"2024-12-24T09:35:48","slug":"how-to-determine-sample-size-in-t-test","status":"publish","type":"post","link":"https:\/\/best-biostatistics.com\/toukei-er\/entry\/how-to-determine-sample-size-in-t-test\/","title":{"rendered":"R \u3067 t \u691c\u5b9a\u306b\u5fc5\u8981\u306a\u30b5\u30f3\u30d7\u30eb\u6570\u3092\u8a08\u7b97\u3059\u308b\u65b9\u6cd5"},"content":{"rendered":"\n

t \u691c\u5b9a\u306e\u30b5\u30f3\u30d7\u30eb\u6570\u8a08\u7b97\u306e\u65b9\u6cd5<\/p>\n\n\n\n\n\n\n\n

t \u691c\u5b9a\u306e\u30b5\u30f3\u30d7\u30eb\u6570\u8a08\u7b97\uff081:1\u306e\u5834\u5408\uff09<\/h2>\n\n\n\n

R \u306e power.t.test()\u3092\u7528\u3044\u308b\u3002<\/p>\n\n\n\n

\u5fc5\u8981\u306a\u60c5\u5831\u306f\u3001\u691c\u51fa\u529b\u3092\u3069\u3046\u3059\u308b\u304b\u3068\u3001\u4e8c\u7fa4\u306e\u5e73\u5747\u5024\u306e\u5dee\u3002<\/p>\n\n\n\n

\u6a19\u6e96\u504f\u5dee\u3067\u5272\u3063\u305f\u6a19\u6e96\u5316\u3055\u308c\u305f\u5dee\u304c\u5fc5\u8981\u3060\u3002<\/p>\n\n\n\n

\u691c\u51fa\u529b\u306f\u6163\u4f8b\u306780%\uff1d0.8\u304c\u591a\u3044\u3002<\/p>\n\n\n\n

90%\uff1d0.9\u3082\u3088\u304f\u3042\u308b\u3002<\/p>\n\n\n\n

\u6a19\u6e96\u5316\u3055\u308c\u305f\u5deed\u304c\u96e3\u3057\u3044\u3002<\/p>\n\n\n\n

\u5148\u884c\u7814\u7a76\u306f\u4e8c\u7fa4\u306e\u5dee\u306e\u6a19\u6e96\u504f\u5dee\u304c\u8a08\u7b97\u3055\u308c\u3066\u3044\u306a\u3044\u3053\u3068\u304c\u307b\u3068\u3093\u3069\u3060\u3002<\/p>\n\n\n\n

\u305d\u3093\u306a\u3068\u304d\u306f\u3001\u4ee5\u4e0b\u306e\u76ee\u5b89\u3092\u4f7f\u3046\u3002<\/p>\n\n\n\n

\u76ee\u5b89<\/td>\u6a19\u6e96\u5316\u3055\u308c\u305f\u5dee\uff08\u52b9\u679c\u91cf\uff09d<\/td><\/tr>
\u5c0f\u3055\u306a\u5dee\u3092\u691c\u51fa\u3057\u305f\u3044\uff08\u5c0f\u3055\u306a\u5dee\u3068\u898b\u7a4d\u3082\u3063\u3066\u3044\u308b\uff09<\/td>0.1 ~ 0.2<\/td><\/tr>
\u4e2d\u304f\u3089\u3044\u306e\u5dee\u3092\u691c\u51fa\u3057\u305f\u3044\uff08\u4e2d\u304f\u3089\u3044\u306e\u5dee\u3068\u898b\u7a4d\u3082\u3063\u3066\u3044\u308b<\/td>0.4 ~ 0.5<\/td><\/tr>
\u5927\u304d\u306a\u5dee\u3092\u691c\u51fa\u3057\u305f\u3044\uff08\u5927\u304d\u306a\u5dee\u3068\u898b\u7a4d\u3082\u3063\u3066\u3044\u308b\uff09<\/td>0.8 ~ 0.9<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n

\u691c\u51fa\u529b80%\u3001\u4e2d\u304f\u3089\u3044\u306a\u5dee\u3092\u691c\u51fa\u3057\u305f\u3044\u3068\u3057\u3066 d=0.5 \u3068\u3059\u308b\u3002<\/p>\n\n\n\n

power.t.test()\u3067\u306f\u3001\u691c\u51fa\u529b\u3092power=\u3067\u3001\u6a19\u6e96\u5316\u3055\u308c\u305f\u5dee\u3092delta=\u3067\u3001\u6307\u5b9a\u3059\u308b\u3002<\/p>\n\n\n\n

power.t.test<\/span>(<\/span>power=<\/span>0.8<\/span>,<\/span>delta=<\/span>0.5<\/span>)<\/span>\n<\/code><\/pre>\n\n\n\n
\u7d50\u679c\u306f\u4e00\u7fa464\u4f8b\u5fc5\u8981\u3068\u8a08\u7b97\u3055\u308c\u308b\u3002<\/p>\n\n\n\n
\u4e8c\u7fa4\u5408\u308f\u305b\u3066128\u4f8b\u304c\u5fc5\u8981\u306a\u30b5\u30f3\u30d7\u30eb\u30b5\u30a4\u30ba\u3060\u3002<\/p>\n\n\n\n
> power.t.test(power=0.8,delta=0.5)\n\n     Two-sample t test power calculation \n\n              n = 63.76576\n          delta = 0.5\n             sd = 1\n      sig.level = 0.05\n          power = 0.8\n    alternative = two.sided\n\nNOTE: n is number in *each* group<\/code><\/pre>\n\n\n\n
t \u691c\u5b9a\u306e\u30b5\u30f3\u30d7\u30eb\u6570\u8a08\u7b97\uff081:1\u306e\u5834\u5408\uff09power.t.test() \u3092\u4f7f\u308f\u306a\u3044\u65b9\u6cd5<\/h2>\n\n\n\n
power.t.test() \u3092\u4f7f\u308f\u306a\u3044\u7c21\u6613\u7684\u306a\u65b9\u6cd5\u306e R \u30b9\u30af\u30ea\u30d7\u30c8\u3092\u4e0b\u8a18\u53c2\u8003\u66f8\u7c4d\u3092\u53c2\u7167\u3057\u3066\u81ea\u4f5c\u3057\u305f\u3002<\/p>\n\n\n\n
t.test.samplesize <- function(sig.level=0.05, power=0.8,\n                              delta, sd=1, alternative=c(\"two.sided\",\"one.sided\"))\n{\n  alternative <- match.arg(alternative)\n  tside <- switch(alternative, one.sided=1, two.sided=2)\n  Za <- qnorm(sig.level\/tside, lower.tail=FALSE)\n  Zb <- qnorm(power)\n  n <- 2*((Za+Zb)\/(delta\/sd))^2\n  NOTE <- \"n is number in *each* group\"\n  METHOD <- \"t-test sample size\"\n  structure(list(n=n,delta=delta, sd=sd,\n                 sig.level=sig.level, power=power,\n                 alternative=alternative, note=NOTE, method=METHOD),\n            class=\"power.htest\")\n}\n<\/code><\/pre>\n\n\n\n
\u7d50\u679c\u306f\u4e00\u7fa463\u4f8b\u3068\u8a08\u7b97\u3055\u308c\u305f\u3002<\/p>\n\n\n\n
power.t.test()\u3092\u4f7f\u3063\u305f\u6642\u306f64\u4f8b\u3068\u8a08\u7b97\u3055\u308c\u305f\u3002<\/p>\n\n\n\n
\u7c21\u6613\u7684\u306a\u65b9\u6cd5\u3067\u3082\u5927\u304d\u306a\u9055\u3044\u304c\u306a\u3044\u3002<\/p>\n\n\n\n
> t.test.samplesize(delta=0.5)\n\n     t-test sample size \n\n              n = 62.79104\n          delta = 0.5\n             sd = 1\n      sig.level = 0.05\n          power = 0.8\n    alternative = two.sided\n\nNOTE: n is number in *each* group<\/code><\/pre>\n\n\n\n
\uff1e\uff1e\u3082\u3046\u7d71\u8a08\u3067\u60a9\u3080\u306e\u306f\u7d42\u308f\u308a\u306b\u3057\u307e\u305b\u3093\u304b\uff1f\u00a0<\/a><\/strong><\/span><\/p>\r\n\"\"<\/a>\r\n
\u21911\u4e07\u4eba\u4ee5\u4e0a\u306e\u533b\u7642\u5f93\u4e8b\u8005\u304c\u8cfc\u8aad\u4e2d<\/span><\/strong><\/span><\/p><\/div>
t \u691c\u5b9a\u306e\u30b5\u30f3\u30d7\u30eb\u6570\u8a08\u7b97\uff08\u5272\u308a\u4ed8\u3051\u6bd4\u5bfe\u5fdc\uff09<\/h2>\n\n\n\n
R \u306e\u6a19\u6e96\u3067\u306f\u30011:2\u3068\u304b1:3\u306b\u5272\u308a\u4ed8\u3051\u308b\u30b5\u30f3\u30d7\u30eb\u30b5\u30a4\u30ba\u8a08\u7b97\u306e\u30d7\u30ed\u30b0\u30e9\u30e0\u306f\u7528\u610f\u3055\u308c\u3066\u3044\u306a\u3044\u3002<\/p>\n\n\n\n
\u81ea\u524d\u3067\u4f5c\u3063\u305f\u30d7\u30ed\u30b0\u30e9\u30e0\u304c\u4ee5\u4e0b\u306e\u901a\u308a\u3002<\/p>\n\n\n\n
power.t.test() \u30671:1\u306e\u30b5\u30f3\u30d7\u30eb\u30b5\u30a4\u30ba\u3092\u8a08\u7b97\u3057\u305f\u5f8c\u3001\u5272\u308a\u4ed8\u3051\u306e\u6bd4(ctrl.ratio)\u3067\u8abf\u6574\u3057\u3066\u3044\u308b\u3002<\/p>\n\n\n\n
power.t.test.unequal <- function(n = NULL, delta = NULL, sd = 1, sig.level = 0.05, power = NULL,\n                                 type = \"two.sample\", alternative = \"two.sided\", strict = FALSE, \n                                 ctrl.ratio = NULL) {\n  res <- power.t.test(n = n, delta = delta, sd = sd, sig.level = sig.level, power = power,\n                      type = type, alternative = alternative, strict = strict)\n  alternative <- match.arg(alternative)\n  tside <- switch(alternative, one.sided=1, two.sided=2)\n  nA <- 2 * res$n \/ (ctrl.ratio + 1)\n  nB <- 2 * res$n * ctrl.ratio \/ (ctrl.ratio + 1)\n  Za <- qnorm(sig.level\/tside, lower.tail=FALSE)\n  power.u <- 1 - pnorm(Za - delta\/sd * (1 \/ sqrt(1 \/ nA + 1 \/ nB)))\n  METHOD <- paste(\"Two-sample\", \"t test power calculation\")\n  structure(list(n.equal = res$n, n.unequal.1 = nA, n.unequal.2 = nB, delta = delta, sd = sd,\n                 sig.level = sig.level, power.equal = res$power, power.unequal = power.u, \n                 alternative = alternative, method = METHOD), class = \"power.htest\")\n}<\/code><\/pre>\n\n\n\n
power\u3068delta\u306f\u5148\u307b\u3069\u3068\u540c\u3058\u306b\u3057\u3066\u30011:1, 1:2, 1:3\u306e\u5834\u5408\u3092\u8a08\u7b97\u3057\u3066\u307f\u308b\u3002<\/p>\n\n\n\n
\u307e\u305a1:1\u306e\u5834\u5408\u3002<\/p>\n\n\n\n
1:1\u306f\u524d\u7bc0\u3068\u540c\u3058\u3088\u3046\u306b\u4e00\u7fa464\u4f8b\u3068\u8a08\u7b97\u3055\u308c\u308b\u3002<\/p>\n\n\n\n
> power.t.test.unequal(power=0.8, delta=0.5, ctrl.ratio=1)\n\n     Two-sample t test power calculation \n\n        n.equal = 63.76576\n    n.unequal.1 = 63.76576\n    n.unequal.2 = 63.76576\n          delta = 0.5\n             sd = 1\n      sig.level = 0.05\n    power.equal = 0.8\n  power.unequal = 0.8060089\n    alternative = two.sided<\/code><\/pre>\n\n\n\n
1:2 \u3068\u3059\u308b\u3068\u300143\u4f8b\u306886\u4f8b\u5fc5\u8981\u3068\u8a08\u7b97\u3055\u308c\u308b\u3002<\/p>\n\n\n\n
\u691c\u51fa\u529b\u306f76%\u306b\u843d\u3061\u308b\u3002<\/p>\n\n\n\n
> power.t.test.unequal(power=0.8, delta=0.5, ctrl.ratio=2)\n\n     Two-sample t test power calculation \n\n        n.equal = 63.76576\n    n.unequal.1 = 42.51051\n    n.unequal.2 = 85.02102\n          delta = 0.5\n             sd = 1\n      sig.level = 0.05\n    power.equal = 0.8\n  power.unequal = 0.7586038\n    alternative = two.sided<\/code><\/pre>\n\n\n\n
1:3 \u3068\u3059\u308b\u3068\u300132\u4f8b\u306896\u4f8b\u5fc5\u8981\u3068\u8a08\u7b97\u3055\u308c\u308b\u3002<\/p>\n\n\n\n
\u691c\u51fa\u529b\u306f\u3055\u3089\u306b\u843d\u3061\u306669%\u306b\u306a\u308b\u3002<\/p>\n\n\n\n
> power.t.test.unequal(power=0.8, delta=0.5, ctrl.ratio=3)\n\n     Two-sample t test power calculation \n\n        n.equal = 63.76576\n    n.unequal.1 = 31.88288\n    n.unequal.2 = 95.64865\n          delta = 0.5\n             sd = 1\n      sig.level = 0.05\n    power.equal = 0.8\n  power.unequal = 0.6861757\n    alternative = two.sided<\/code><\/pre>\n\n\n\n
\u3053\u3053\u304b\u3089\u306f1:1\u304c\u6700\u3082\u691c\u51fa\u529b\u304c\u9ad8\u3044\u3053\u3068\u304c\u308f\u304b\u308b\u3002<\/p>\n\n\n\n
1:1\u306e\u5272\u308a\u4ed8\u3051\u304c\u304a\u3059\u3059\u3081\u306a\u306e\u306f\u3001\u691c\u51fa\u529b\u304c\u9ad8\u3044\u304b\u3089\u3060\u3002<\/p>\n\n\n\n
1:1\u5272\u308a\u4ed8\u3051\u51fa\u306a\u3044\u5834\u5408\u3001\u6700\u7d42\u7684\u306a\u691c\u51fa\u529b\u309280%\u306b\u4fdd\u3064\u305f\u3081\u306b\u306f\u6700\u521d\u306e\u691c\u51fa\u529b\u3092\u4e0a\u3052\u3066\u304a\u304b\u306d\u3070\u306a\u3089\u306a\u3044\u3002<\/p>\n\n\n\n
\u6700\u521d\u306e\u691c\u51fa\u529b\u309285%\u306b\u3057\u3066\u304a\u3044\u3066\u30011:2\u5272\u308a\u4ed8\u3051\u306e\u30b5\u30f3\u30d7\u30eb\u30b5\u30a4\u30ba\u3092\u8a08\u7b97\u3059\u308b\u3068\u3001\u3069\u3046\u306a\u308b\u304b\uff1f<\/p>\n\n\n\n
> power.t.test.unequal(power=0.85, delta=0.5, ctrl.ratio=2)\n\n     Two-sample t test power calculation \n\n        n.equal = 72.80053\n    n.unequal.1 = 48.53369\n    n.unequal.2 = 97.06738\n          delta = 0.5\n             sd = 1\n      sig.level = 0.05\n    power.equal = 0.85\n  power.unequal = 0.8116907\n    alternative = two.sided<\/code><\/pre>\n\n\n\n
\u7d50\u679c\u306f\u300149\u4f8b\u306898\u4f8b\u3067\u3001\u6700\u7d42\u306e\u691c\u51fa\u529b\u304c81\uff05\u306b\u306a\u308b\u3002<\/p>\n\n\n\n
\u3053\u308c\u306a\u3089\u3070\u3001\u5fc5\u8981\u75c7\u4f8b\u6570\u306f64\u4f8b\u304b\u308949\u4f8b\u306b15\u4f8b\u6e1b\u308a\u3001\u691c\u51fa\u529b\u306f80%\u3092\u4fdd\u3063\u3066\u3044\u308b\u305f\u3081\u3001\u6709\u7528\u304b\u3064\u9069\u5207\u306a\u30c7\u30b6\u30a4\u30f3\u5909\u66f4\u3060\u3002<\/p>\n\n\n\n
\u305f\u3060\u3057\u3001\u5408\u8a08\u306e\u5fc5\u8981\u4f8b\u6570\u4f8b\u6570\u306f49\uff0b98\uff1d147\u3068\u306a\u308a\u30011:1\u306e\u3068\u304d\u306e\u5408\u8a08\u4f8b\u6570128\u4f8b\u3088\u308a19\u4f8b\u5897\u52a0\u3059\u308b\u3002<\/p>\n\n\n\n
1:3\u306e\u5834\u5408\u306f\u30011:1\u306e\u691c\u51fa\u529b\u309290%\u306b\u3059\u308b\u3068\u30011:3\u306e\u691c\u51fa\u529b\u304c80%\u3092\u8d85\u3048\u308b\u3002<\/p>\n\n\n\n
> power.t.test.unequal(power=0.90, delta=0.5, ctrl.ratio=3)\n\n     Two-sample t test power calculation \n\n        n.equal = 85.03129\n    n.unequal.1 = 42.51564\n    n.unequal.2 = 127.5469\n          delta = 0.5\n             sd = 1\n      sig.level = 0.05\n    power.equal = 0.9\n  power.unequal = 0.8060558\n    alternative = two.sided<\/code><\/pre>\n\n\n\n
\u5fc5\u8981\u4f8b\u6570\u306f43\u4f8b\u3068128\u4f8b\u3067\u3001\u5408\u8a08171\u4f8b\u5fc5\u8981\u306b\u306a\u308b\u30021:1\u306e\u6642\u3088\u308a43\u4f8b\u5897\u52a0\u3059\u308b\u3002<\/p>\n\n\n\n
\u75c7\u4f8b\u3092\u96c6\u3081\u308b\u306e\u304c\u5927\u5909\u306a\u5834\u5408\u306f\u81f4\u3057\u65b9\u306a\u3044\u304c\u3001\u3067\u304d\u308c\u30701:1\u306b\u3057\u305f\u307b\u3046\u304c\u3001\u5408\u8a08\u306e\u5fc5\u8981\u306a\u30b5\u30f3\u30d7\u30eb\u30b5\u30a4\u30ba\u306f\u5c0f\u3055\u304f\u306a\u308b\u3002<\/p>\n\n\n\n
t \u691c\u5b9a\u306e\u30b5\u30f3\u30d7\u30eb\u6570\u8a08\u7b97\uff08\u5272\u308a\u4ed8\u3051\u6bd4\u5bfe\u5fdc\uff09power.t.test() \u3092\u4f7f\u308f\u306a\u3044\u65b9\u6cd5<\/h2>\n\n\n\n
1:1\u306e\u5834\u5408\u3068\u540c\u69d8\u306b\u3001power.t.test() \u3092\u4f7f\u308f\u306a\u3044\u7c21\u6613\u7684\u306a\u65b9\u6cd5\u306e\u30b9\u30af\u30ea\u30d7\u30c8\u3092\u4f5c\u6210\u3057\u3066\u3001power.t.test() \u3092\u4f7f\u3063\u305f\u5834\u5408\u3068\u6bd4\u8f03\u3057\u305f\u3002<\/p>\n\n\n\n
power.t.test.unequal.approx <- function(\n    delta = NULL, sd = 1, sig.level = 0.05, power = 0.8,\n    alternative = c(\"two.sided\",\"one.sided\"),\n    ctrl.ratio = NULL) {\n  alternative <- match.arg(alternative)\n  tside <- switch(alternative, one.sided=1, two.sided=2)\n  Za0 <- qnorm(sig.level\/tside, lower.tail=FALSE)\n  Zb0 <- qnorm(power)\n  n <- 2*((Za0+Zb0)\/(delta\/sd))^2\n  nA <- 2 * n \/ (ctrl.ratio + 1)\n  nB <- 2 * n * ctrl.ratio \/ (ctrl.ratio + 1)\n  Za <- qnorm(sig.level\/tside, lower.tail=FALSE)\n  power.u <- 1 - pnorm(Za - delta\/sd * (1 \/ sqrt(1 \/ nA + 1 \/ nB)))\n  METHOD <- paste(\"Two-sample\", \"t test power calculation\")\n  structure(list(n.equal = n, n.unequal.1 = nA,\n                 n.unequal.2 = nB, delta = delta, sd = sd,\n                 sig.level = sig.level, power.equal = power,\n                 power.unequal = power.u,  alternative = alternative,\n                 method = METHOD), class = \"power.htest\")\n}\n<\/code><\/pre>\n\n\n\n
\u307e\u305a1:1\u306e\u5834\u5408\u3002<\/p>\n\n\n\n
power.t.test()\u306e\u7d50\u679c\u3068\u6bd4\u3079\u3001\u5404\u7fa41\u4f8b\u5c11\u306a\u3044\u306e\u306f\u4e0a\u8ff0\u3057\u305f\u901a\u308a\u3002<\/p>\n\n\n\n
> power.t.test.unequal.approx(power=0.8, delta=0.5, ctrl.ratio=1)\n\n     Two-sample t test power calculation \n\n        n.equal = 62.79104\n    n.unequal.1 = 62.79104\n    n.unequal.2 = 62.79104\n          delta = 0.5\n             sd = 1\n      sig.level = 0.05\n    power.equal = 0.8\n  power.unequal = 0.8\n    alternative = two.sided<\/code><\/pre>\n\n\n\n
1:2\u306e\u5834\u5408\u306f\u30011:1\u306e\u5834\u5408\u306e\u691c\u51fa\u529b\u309285%\u306b\u3059\u308b\u3068\u3001\u691c\u51fa\u529b\u304c80%\u3092\u8d85\u3048\u308b\u306e\u306f\u540c\u3058\u3002<\/p>\n\n\n\n
> power.t.test.unequal.approx(power=0.85, delta=0.5, ctrl.ratio=2)\n\n     Two-sample t test power calculation \n\n        n.equal = 71.82718\n    n.unequal.1 = 47.88479\n    n.unequal.2 = 95.76957\n          delta = 0.5\n             sd = 1\n      sig.level = 0.05\n    power.equal = 0.85\n  power.unequal = 0.8064989\n    alternative = two.sided<\/code><\/pre>\n\n\n\n
\u5c0f\u3055\u3044\u4f8b\u6570\u306e\u7fa4\u306f49\u304b\u308948\u4f8b\u306b1\u4f8b\u6e1b\u5c11\u3001\u5927\u304d\u3044\u4f8b\u6570\u306e\u7fa4\u306f98\u4f8b\u304b\u308996\u4f8b\u306b2\u4f8b\u6e1b\u5c11\u3057\u305f\u3002<\/p>\n\n\n\n
\u5927\u304d\u306a\u9055\u3044\u306f\u306a\u3044\u3068\u8a00\u3048\u308b\u3002<\/p>\n\n\n\n
1:3\u306e\u5834\u5408\u306f\u30011:1\u306e\u5834\u5408\u306e\u691c\u51fa\u529b\u309290%\u306b\u3059\u308b\u3068\u3001\u691c\u51fa\u529b\u304c80%\u3092\u8d85\u3048\u308b\u306e\u306f\u3053\u3061\u3089\u3082\u540c\u69d8\u3002<\/p>\n\n\n\n
> power.t.test.unequal.approx(power=0.90, delta=0.5, ctrl.ratio=3)\n\n     Two-sample t test power calculation \n\n        n.equal = 84.05938\n    n.unequal.1 = 42.02969\n    n.unequal.2 = 126.0891\n          delta = 0.5\n             sd = 1\n      sig.level = 0.05\n    power.equal = 0.9\n  power.unequal = 0.8015779\n    alternative = two.sided<\/code><\/pre>\n\n\n\n
\u5c0f\u3055\u3044\u4f8b\u6570\u306e\u7fa4\u306f\u300143\u4f8b\u304b\u3089\u5909\u5316\u306a\u3057\u3002<\/p>\n\n\n\n
\u5927\u304d\u3044\u4f8b\u6570\u306e\u7fa4\u306f\u3001128\u4f8b\u304b\u3089127\u4f8b\u306b1\u4f8b\u6e1b\u5c11\u3002<\/p>\n\n\n\n
\u3053\u3061\u3089\u3082\u5927\u304d\u306a\u9055\u3044\u306f\u306a\u3044\u3068\u8a00\u3048\u308b\u3002<\/p>\n\n\n\n
\u3088\u3063\u3066\u3001\u7c21\u6613\u7684\u306a\uff08\u8fd1\u4f3c\u7684\u306a\uff09\u65b9\u6cd5\u306f\u3001power.t.test() \u3092\u4f7f\u3063\u305f\u65b9\u6cd5\u3068\u3001\u5b9f\u7528\u4e0a\u306f\u5dee\u304c\u306a\u3044\u3068\u8a00\u3048\u308b\u3060\u308d\u3046\u3002<\/p>\n\n\n\n
t \u691c\u5b9a\u306e\u30b5\u30f3\u30d7\u30eb\u6570\u8a08\u7b97\uff08\u5272\u308a\u4ed8\u3051\u6bd4\u5bfe\u5fdc\uff09power.t.test() \u3092\u4f7f\u308f\u306a\u3044\u3082\u3046\u5c11\u3057\u7c21\u5358\u306a\u65b9\u6cd5<\/h2>\n\n\n\n
EZR \u306e\u300c2\u7fa4\u306e\u5e73\u5747\u5024\u306e\u6bd4\u8f03\u306e\u305f\u3081\u306e\u30b5\u30f3\u30d7\u30eb\u30b5\u30a4\u30ba\u306e\u8a08\u7b97\u300d\u3092\u53c2\u8003\u306b\u3057\u3066\u3001\u3082\u3046\u5c11\u3057\u7c21\u5358\u306b\u5272\u308a\u4ed8\u3051\u6bd4\u3042\u308a\u306e\u30b5\u30f3\u30d7\u30eb\u30b5\u30a4\u30ba\u8a08\u7b97\u304c\u3067\u304d\u308b\u3088\u3046\u306b\u6539\u9020\u3057\u305f\u3002<\/p>\n\n\n\n
Case\u306e\u4f8b\u6570 n1 \u3092\u8a08\u7b97\u3057\u305f\u3068\u304d\u306b\u5207\u308a\u4e0a\u3052\u3066\u3001\u305d\u3053\u306b\u5272\u308a\u4ed8\u3051\u6bd4\u3092\u304b\u3051\u308b\u3088\u3046\u306b\u5909\u66f4\u3057\u305f\u3002<\/p>\n\n\n\n
t.test.samplesize.unequal.approx2 <- function(\n    sig.level=0.05, power=0.8,\n    delta, sd=1, alternative=c(\"two.sided\",\"one.sided\"),ctrl.ratio=NULL\n    )\n{\n  alternative <- match.arg(alternative)\n  tside <- switch(alternative, one.sided=1, two.sided=2)\n  Za <- qnorm(sig.level\/tside, lower.tail=FALSE)\n  Zb <- qnorm(power)\n  n1 <- ceiling((1+1\/ctrl.ratio)*((Za+Zb)\/(delta\/sd))^2)\n  n2 <- n1*ctrl.ratio\n  NOTE <- \"n1 or n2 is number in *each* group\"\n  METHOD <- \"t-test sample size\"\n  structure(list(n1=n1, n2=n2, \"control ratio\"=ctrl.ratio, delta=delta, sd=sd,\n                 sig.level=sig.level, power=power,\n                 alternative=alternative, note=NOTE, method=METHOD),\n            class=\"power.htest\")\n}\n<\/code><\/pre>\n\n\n\n
sig.level=0.05, power=0.8, sd=1\u306e\u30c7\u30d5\u30a9\u30eb\u30c8\u306e\u307e\u307e\u3068\u3057\u3066\u3001delta=0.5, ctrl.ratio=1,2,3\u306e\u3068\u304d\u3001\u7d50\u679c\u306f\u4ee5\u4e0b\u306e\u901a\u308a\u3002<\/p>\n\n\n\n
> t.test.samplesize.unequal.approx2(delta=0.5,ctrl.ratio=1)\n\n     t-test sample size \n\n             n1 = 63\n             n2 = 63\n  control ratio = 1\n          delta = 0.5\n             sd = 1\n      sig.level = 0.05\n          power = 0.8\n    alternative = two.sided\n\nNOTE: n1 or n2 is number in *each* group\n\n\n> t.test.samplesize.unequal.approx2(delta=0.5,ctrl.ratio=2)\n\n     t-test sample size \n\n             n1 = 48\n             n2 = 96\n  control ratio = 2\n          delta = 0.5\n             sd = 1\n      sig.level = 0.05\n          power = 0.8\n    alternative = two.sided\n\nNOTE: n1 or n2 is number in *each* group\n\n\n> t.test.samplesize.unequal.approx2(delta=0.5,ctrl.ratio=3)\n\n     t-test sample size \n\n             n1 = 42\n             n2 = 126\n  control ratio = 3\n          delta = 0.5\n             sd = 1\n      sig.level = 0.05\n          power = 0.8\n    alternative = two.sided\n\nNOTE: n1 or n2 is number in *each* group\n<\/code><\/pre>\n\n\n\n
t \u691c\u5b9a\u306e\u30b5\u30f3\u30d7\u30eb\u6570\u8a08\u7b97\u3092\u30a8\u30af\u30bb\u30eb\u3067<\/h2>\n\n\n\n
\u30a8\u30af\u30bb\u30eb\u30d5\u30a1\u30a4\u30eb\u8ca9\u58f2\u4e2d\u3002<\/p>\n\n\n\n
\u3088\u3051\u308c\u3070\u3069\u3046\u305e\u3002<\/p>\n\n\n\n
t\u691c\u5b9a \u5e73\u5747\u5024\u306e\u5dee\u306e\u691c\u5b9a\uff08\u5272\u308a\u4ed8\u3051\u6bd4\u306b\u3082\u5bfe\u5fdc\uff09\u30b5\u30f3\u30d7\u30eb\u30b5\u30a4\u30ba\u8a08\u7b97\u3010\u6539\u826f\u7248\u3011\u3010\u30a8\u30af\u30bb\u30eb\u3067\u30b5\u30f3\u30d7\u30eb\u30b5\u30a4\u30ba\u3011 | TKER SHOP<\/a><\/p>\n\n\n\n
\u4f7f\u3044\u65b9\u89e3\u8aac\u52d5\u753b\u3002<\/p>\n\n\n\n
\u3053\u3061\u3089\u3082\u3088\u3051\u308c\u3070\u3001\u305c\u3072\u3002<\/p>\n\n\n\n