Professional Documents
Culture Documents
Take a look at the first box in your output file called Group Statistics. In the first column, you will see the number 1 and the number 2 under the word IVS G!". These are the numbers that we chose to represent our two IV conditions, su#ar $1% and no su#ar $2%. &ou can find out some descripti'e statistics about each condition by readin# across each row in this box. &ou can also see the number of participants per condition.
Example
In the Group Statistics box, the mean for condition 1 $su#ar% is (.2). The mean for condition 2 $no su#ar% is 2.2). The standard de'iation for condition 1 is 1.*) and for condition 2, ).+(. The number of participants in each condition $,% is -.
Example
.e can see how many data points were entered for each condition. If we know that we had - participants per condition in our experiment, but , 0 ( for condition 1 on this printout, this would be an indication that we had not entered all of the participant data in our data file. The condition means are also 'ery important. They show us the ma#nitude of the difference between conditions and we can see which #roup has a hi#her mean. 1or example, we can see that the mean for condition1 is almost twice that of condition 2. .e can see that participants in the su#ar condition are rememberin# nearly twice the amount of words when compared to the no su#ar condition.
This is the next box you will look at. !t first #lance, you can see a lot of information and that mi#ht feel intimidatin#. 2ut don3t worry, you actually only ha'e to look at half of the information in this box, either the top row or the bottom row.
+ur Example The Si#. $2/Tailed% 'alue in our example is ).)2. This 'alue is less than .)-. 2ecause of this, we can conclude that there is a statistically si#nificant difference between the mean number of words recalled for the su#ar and no su#ar conditions. Since our Group Statistics box re'ealed that the 9ean for the su#ar condition was #reater than the 9ean for the no su#ar condition, we can conclude that participants in the su#ar condition were able to recall si#nificantly more words than participants in the no su#ar condition.
-earsons r &ou can find the 8earson3s r statistic in the top of each box. The 8earson3s r for the correlation between the water and skin 'ariables in our example is ).=+-.
When -earsons r is close to .% This means that there is a stron# relationship between your two 'ariables. This means that chan#es in one 'ariable are stron#ly correlated with chan#es in the second 'ariable. In our example, 8earson3s r is ).=+-. This number is 'ery close to 1. 1or this reason, we can conclude that there is a stron# relationship between our water and skin 'ariables. >owe'er, we cannot make any other conclusions about this relationship, based on this number. When -earsons r is close to #% This means that there is a weak relationship between your two 'ariables. This means that chan#es in one 'ariable are not correlated with chan#es in the second 'ariable. If our 8earson3s r were ).)1, we could conclude that our 'ariables were not stron#ly correlated. When -earsons r is positive '/*% This means that as one 'ariable increases in 'alue, the second 'ariable also increase in 'alue. Similarly, as one 'ariable decreases in 'alue, the second 'ariable also decreases in 'alue. This is called a positi'e correlation. In our example, our 8earson3s r 'alue of ).=+- was positi'e. .e know this 'alue is positi'e because S8SS did not put a ne#ati'e si#n in front of it. So, positi'e is the default. Since our example 8earson3s r is positi'e, we can conclude that when the amount of water increases $our first 'ariable%, the participant skin elasticity ratin# $our second 'ariable% also increases. When -earsons r is ne!ative ')*%
This means that as one 'ariable increases in 'alue, the second 'ariable decreases in 'alue. This is called a ne#ati'e correlation. In our example, our 8earson3s r 'alue of ).=+- was positi'e. 2ut what if S8SS #enerated a 8earson3s r 'alue of /).=+-< If S8SS #enerated a ne#ati'e 8earson3s r 'alue, we could conclude that when the amount of water increases $our first 'ariable%, the participant skin elasticity ratin# $our second 'ariable% decreases. Si! '()Tailed* value &ou can find this 'alue in the :orrelations box. This 'alue will tell you if there is a statistically si#nificant correlation between your two 'ariables. In our example, our Si#. $2/tailed% 'alue is ).))2.
If the Si! '()Tailed* value is !reater than #$% &ou can conclude that there is no statistically si#nificant correlation between your two 'ariables. That means, increases or decreases in one 'ariable do not si#nificantly relate to increases or decreases in your second 'ariable. If the Si! '()Tailed* value is less than or equal to "#$% &ou can conclude that there is a statistically si#nificant correlations between your two 'ariables. That means, increases or decreases in one 'ariable do si#nificantly relate to increases or decreases in your second 'ariable. +ur Example The Si#. $2/Tailed% 'alue in our example is ).))2. This 'alue is less than . )-. 2ecause of this, we can conclude that there is a statistically si#nificant correlation between amount of water consumed in #lasses and participant ratin# of skin elasticity. Warnin! a0out the Si! '()Tailed* value .hen you are computin# 8earson3s r, si#nificance is a messy topic. .hen you ha'e small samples, for example only a few participants, moderate correlations may misleadin#ly not reach si#nificance. .hen you ha'e lar#e samples, for example many participants, small correlations may misleadin#ly turn out to be si#nificant. Some researchers think that
si#nificance should be reported but perhaps should recei'e less focus when it comes to 8earson3s r. So &hat a0out the scatterplot1 &ou can find your scatterplot in your output file. It will look somethin# like the #raph below. &ou will see a bunch of dots. &our scatterplot can tell you about the relationship between 'ariables, ?ust like 8earson3s r. .ith it, you can determine the stren#th and direction of the relationship between 'ariables.
2elationship stren!th Try to ima#ine a line that connects the dots in your scatterplot. Is this an easy or difficult task< This task can help you determine the stren#th of the relationship between your two 'ariables. If your 'ariables ha'e a stron# relationship, it will be easy for your to ima#ine a line connectin# all of the dots. 1or example, in our example scatterplots, the dots seem to #o to#ether to form a strai#ht line. >owe'er, some scatterplots do not look like this. .ith some scatterplots, the dots are scattered about so that it is 'ery hard to ima#ine a line connectin# them. The dots are not densely positioned in one place. Instead, they are all o'er the place. .hen this is the case, your 'ariables may not ha'e a stron# relationship. 2elationship 3irection &ou can use your scatterplot to understand the direction of your relationship. &our scatterplot can tell you if you ha'e a positi'e, ne#ati'e or @ero correlation. -ositive correlation in a scatterplot If the line that you ima#ine in your #raph slopes upward from @ero, you can conclude that you ha'e a positi'e correlation between your 'ariables. Increases in one 'ariable are correlated with increases in your other 'ariable. Similarly, decreases in one 'ariable are correlated with decreases in your other 'ariable.
4e!ative correlation in a scatterplot If the line that you ima#ine in your #raph starts hi#h at @ero and #radually slopes downward, you can conclude that you ha'e a ne#ati'e correlation between your 'ariables. Increases in one 'ariable are correlated with decreases in your other 'ariable. 5ero correlation in a scatterplot If the line that you ima#ine does not slop, or you can3t ima#ine a line at all, you can conclude that you ha'e a @ero correlation between your 'ariables. That means that your 'ariables are not related to one another. Increases or
decreases in one 'ariable ha'e no effect on increases or decreases in your second 'ariable.