# Help with my Editing Test Survey

#### EngPhDGuy

##### New Member
Hello. I'm a PhD student in English, and I need some direction. I have taken a quant methods course but it appears I need a refresher.

Let me give you a brief summary of my project: I sent a survey out to several business execs asking them to rank (on a 3-point Likert scale) the probability of 15 grammar/style errors showing up on the editing tests they writer/administer to new employees. The Likert scale reads: 1 = Sure Bet 2 = Likely to Appear and 3 = Little to No Chance

To muddy the waters further, this section of my survey had a total 53 participants; however, participants but were asked not rank specific grammar/style error they were unfamiliar with. (I have attached some raw data to this post with some of this info.)

Here's what I want to do: I want to divide this data into 3 statistically sound hierarchies/levels (i.e., sure bet to be included in an editing test, most likely to occur, and little to know chance of occurring), but I'm not sure what statistical test to use to separate the variables. For example, “fragment” (1.4) would be placed in the first hierarchy/level, but what test do I run to determine if “comma in a compound sentence” (1.6) goes in that first hierarchy or gets bumped to the second? I know there's a test to determine if the "space" between 1.4 and 1.6 is statistically significant, but I don't know what it would be.

Likewise, since these numbers are so close together can I even claim that “fragment” (1.4) would be in the first/"sure bet" category"?

Finally, I'm not sure how to account for every variable having a different response rate. (I know SPSS has some control for that, but again, not sure what/where it is)

Hope this makes sense. I'm not asking any one to do my work for me, but any suggestions would be much appreciated.

-Ryan

#### Tanya Silva

##### New Member
Hi
How did you calclulate the average?
Do you use SPSS or different stat package?

#### EngPhDGuy

##### New Member
The averages were calcuated by the online survey softare I use; however, I also work in SPSS.

#### Tanya Silva

##### New Member
well, I am not a pro in stats, but I cannot get what does this average actually mean. 1.9 ( first row) does not make sence, 42% said it does bother them, 28% said likely, ect.
But that;s not my point, you are using discrete data and therefore there is no averages, but proportions. You can use PIVOT table to organize your data with graph output.

#### EngPhDGuy

##### New Member
Thanks, Tonya.

I'm not looking for level of bothersome though. This portion of the survey asks respondents to rate the probability of 15 grammar/style errors appearing on the editing tests they writer/administer to new employees. The Likert scale reads: 1 = Sure Bet 2 = Likely to Appear and 3 = Little to No Chance.

So do I need to export my data and separate it by individual response?

-Ryan

#### Tanya Silva

##### New Member
Hey
As I understand all you want to do is to show what people think about those mistakes. If I would such a research I would not go deep in calculations. I cannot even think about any Ho you could use ( but as I said I am not a pro at all).
All I would do is to separate percentages and numbers for each respond. Then put data in PIVOT table , and based on results conclude something like: 1. it appears that business people pay more attention to blah-blah than to blah blah. It is used to be believed ( if you have that info) that business people pay more attention to blah blah. However, from my study we can see that it is questionable. And so on…You can create bar charts for each outcome, so you have nice pics to support your words.