Can someone please explain chi-square , t test and anova ? I can memorize whats given in FA but I doubt that will be enough to answer tricky questions about it unless I actually understand what's going on.
I got this Somewhere from the internet and Bookmarked it for later:
I think it's important to know these two first;
1-"Interval variable" (values from where we can drive their "Mean")
e.g. measurements of B.P, Weight, Cholesterol levels etc
2-"Nominal variable" ( different categories or "GROUPS")
e.g.
*Gender ---is a Nominal that has Two Groups (Male and Female)
*AGE-------is a Nominal that we can have either TWO groups (Elderly and Middle age) or
THREE Groups(Elderly, Middle-age and Children) etc
*Race------a Nominal that can have many groups
Two groups (African Americans and Latino)
Three groups(African Americans, Latino and Asian) and so forth
e.g.
*"GENDER"(Men and Women)....................................................(Single Nominal)
*"GENDER"(Men and Women) n "RACE"(Caucasian and African)....(Two Nominals )
e.g we compare Blood Pressure measurements
*INTERVAL variable ("MEANS" of B.P measurements)
*NOMINAL variable ("GROUPS" )
so to compare MEANS of B.P. b/w two are more groups we do these tests
t-Test:
1 Interval (Blood pressure Means)
1 Nominal with "ONLY TWO GROUPS"
e.g. comparing the BP means b/w (Latinos) and (African Americans)
*for a t-Test it will be a wrong statement to say
e.g comparing B.P means b/w (Latinos), (African Americans) and (Asians)....*got 3 groups
Now what's the difference B/W t-Test and ANOVA
t-Test------>ONLY TWO Groups comparison
ANOVA------------>TWO Groups or more than Two Groups comparison
NOTE:these two will give u the Identical results if comparing "TWO Groups"
t-Test................... 1 Interval and 1 Nominal with TWO Groups
One-way ANOVA.....1 Interval and 1 Nominal with TWO Groups
and for the above reason they rather use One-way ANOVA when more than two groups are there to be compared e.g as above
comparing B.P means b/w (Latinos), (African Americans) and (Asians).......3 groups
Two-way ANOVA:
1 Interval
2 Nominal e.g. Gender (Men and Women) *AND Race (Caucasians and African Americans)
---->we do these tests to see If there is a Difference b/w groups ?
t-statistics (t-Test) and f-statistics(ANOVA) will be compared to get a p-value
If p-value = 0.05 or less -----it is said to be "Statistically Significant' and we say that there is a Difference in these Groups.
Chi-square:
we compare b/w "TWO NOMINAL" (with ANY # of GROUPS)
(Note; we are not using any Interval variable here , just the NOMINAL data)
best example...."Testing Drug Efficacy"
First Nominal........(New drug group) and (Placebo group)
Second Nominal ...(Recovered group) and(Not recovered group)
***I think this is all u have to do is recognize what kind of data is presented , how many Nominals, How many groups and If Interval variable is used or not.
1 Interval and 1 Nominal (Two Groups).......................t-Test or One-way ANOVA
1 Interval and 1 Nominal (More than Two groups)........One-way ANOVA
1 Interval and *2 Nominal (many groups)....................Two-way ANOVA
.....................*2 Nominal (any# of Groups).................Chi-square
Good Luck
