Placeholder Image

字幕列表 影片播放

  • Let's say you want to find out

    假設你想了解

  • if the beverage that people drink

    一般人喝下飲料後

  • affects their reaction time.

    在體內平均的反應時間

  • So you set up an experiment with three groups of people.

    因此你設計一個有三組人的實驗

  • The first group gets water to drink.

    第一組獲得水喝

  • The second group gets some sugary fruit juice,

    第二組獲得含糖果汁

  • and the third group gets coffee.

    而第三組則獲得咖啡

  • Now you test everyone's reaction time.

    現在你來測試大家的反應時間

  • And you want to know if there's any difference

    而你想知道的是,其間有任何差異

  • in reaction time between the groups.

    三組之間的反應時間

  • The null hypothesis says that the mean reaction time

    因此先建立虛無假設平均反應時間

  • for all three groups is the same.

    三組都相同

  • If there were only two groups,

    如果只有兩組的情況下

  • you could use a t-test to find out

    你可以使用 t檢定來找出

  • if there's a difference between them.

    其間所存在的差異

  • But when you have three groups or more,

    但如果你有三組或三組以上

  • you need to use a different approach--

    你必須使用另一種不同的方法--

  • the analysis of variance.

    變異數分析

  • When you do the experiment,

    當你開始實驗

  • the scores won't all be the same.

    所有的分數都不一致

  • The total variation of all the scores

    所有分數的總變異量

  • is made up of two parts:

    都由兩部分所構成

  • The variation within each group,

    各組內的變動

  • because the people in each group

    因為每一組內的人

  • have different reaction times,

    都有不同的反應時間

  • and the variation between the groups,

    加上組與組之間的變異

  • because the drinks you gave each group are different.

    由於你給每一組的飲料都不一樣

  • Here's an example.

    以下即是一例

  • Look at this set of scores.

    來看一下這一組的分數

  • They've been sorted into order

    它們已照順序排列

  • to make it easier to see the patterns.

    以易於看出其間的模式

  • You can see that there's a lot of variation

    你可以看到它們的差異很大

  • within each group;

    在各組之內

  • some people are faster and some are much slower.

    有些人很快,有些人則很慢

  • But all the groups look pretty much alike;

    但各組看起都很相像

  • there's not much variation between the groups.

    雖然組與組之間差異不大

  • In this case, you'd say that most of the difference

    在這個案例中,你只能說最大的差異

  • is due to the people, and the drink

    來自於人,至於飲料

  • didn't make much of a difference.

    並沒有產生很大的差異

  • You can't reject the null hypothesis;

    你無法拒絕虛無假設

  • which is that the type of drink doesn't have

    主要是這類的飲料

  • any effect on reaction time.

    在反應時間上並沒有任何效果

  • Now let's look at a different set of numbers.

    現在,我們來看一下不同組的數字

  • In this case, all the scores within each group

    在這個案例中,各組內的所有分數

  • are very close to one another.

    彼此非常接近

  • There's not a lot of variance within each group,

    各組內的變異不大

  • but the groups are very different from one another.

    但組與組之間差異卻很大

  • There's a lot of difference between the groups.

    使得各組之間差異就變大

  • In this case, you would reject the null hypothesis.

    在這個案例子,你可以拒絕虛無假設

  • In this case, the type of drink makes a big difference.

    在此案例中,飲料的種料造成差異

  • So here's the idea behind analysis of variance:

    因此,在變異數分析之後的想法是

  • Figure out how much of the total variance comes from the between-groups variance

    找出來自組與組之間的變化總量

  • and the within-groups variance.

    以及組內的變化量

  • Take the ratio of between-groups

    再計算組間

  • to within-groups variance,

    和組內變異量的比例

  • and the larger this number is,

    比例值越大

  • the more likely it is

    它越有可能

  • that the means of the groups really *are* different,

    組與組的平均數真的「是」有些差異

  • and that you should reject the null hypothesis.

    而依此,你就可以拒絕先前設定的虛無假設

  • In the examples, it was obvious where the variance was.

    在這個例子中,變異數非常明顯

  • Now look at these numbers.

    現在看看這些數字

  • You probably can't tell

    你可能無法分辨

  • if there's a significant effect

    其中有任何顯著效果

  • because it's not clear whether there's

    因為其中並不明顯

  • more variance within groups or between groups,

    在各組內或組與組間,是否有較多的變異

  • or how much.

    或是變異有多大

  • The calculations show that the ratio is 4.27,

    計算結果顯示比率為4.27

  • which has a probability of .04,

    機率是0.04

  • so in this case, you can reject the null hypothesis.

    因此在這個例子中,可以拒絕虛無假設

  • With these numbers, the drink you give the people

    以這些數字來判斷,你給人們喝的飲料

  • does have an effect on their reaction time.

    在他們的反應時間確實有效

  • What's that "2,12" doing there?

    那這裡「2,12」要做什麼呢?

  • Those are the degrees of freedom

    它們是所謂的自由度

  • for variance between groups

    即組間的變異

  • and variance within groups.

    和組內的變異

  • And here's how you calculate the degrees of freedom

    此處你該如何計算自由度

  • when you report results for analysis of variance.

    當你要報告變異數的結果

  • This trick of separating the variance

    這是分離變異數的小技巧

  • not only when you have three or more groups,

    不管在你有三組或三組以上

  • it also works when you have multiple variables.

    或是有更多重的變異數時

  • For example, if you test three groups

    例如,你要檢定三組

  • for reaction time in the morning,

    在清晨時的反應時

  • and you test another three groups in the evening,

    或是檢定另外三組在傍晚的反應時間

  • an analysis of variance can tell you

    變異數分析就能告訴你

  • if there's a significant effect

    其間是否顯著

  • for the type of drink,

    對飲料的種類

  • or if the time of day makes a difference,

    或者是因每天時段不同差生的差異

  • or if there's some interaction.

    又或者是反應本身

  • For example, coffee might be more effective in the morning than in the evening.

    例如,咖啡本身在清晨時段就比較傍晚強烈

  • So to recap, here's the main idea of analysis of variance:

    來回顧一下,這裡變異數的主要想法是

  • You figure hοw much of the total variance

    你已體會到總變異量的多寡

  • comes from between the groups,

    來自於組間

  • and how much comes from within the groups.

    以及有多少來自於組內

  • If most of the variation is between groups,

    如果大部分的變異來自於組間

  • there's probably a significant effect;

    造成顯著的機率比較高

  • if most of the variation is within groups,

    如果大部分的變異來自於組內

  • there's probably not a significant effect.

    顯著的機率就會降低

Let's say you want to find out

假設你想了解

字幕與單字

單字即點即查 點擊單字可以查詢單字解釋