The title of this blog sounds very mysterious, doesn’t it? Secrets of statistics – they do feel a bit secretive at times. But, like anything, as soon as you know how to do it, it’s no longer a secret. Statistics can be fun, and most importantly, they help us finish our projects and make the impact on the world that we’ve been dreaming of.
So, to help you see that, I’ve decided to share the formula I follow with every one of my 1:1 statistics clients.
This process has helped so many researchers go from feeling stuck and unsure of what to do next, to feeling in control and accomplished as they complete their paper.
Thing is, there isn’t really a big secret that I unveil on this programme. It’s kind of the opposite actually: together we go right back to basics and start from there.
The basics are soooo important
When people come to me for help with their analysis it’s usually because they’ve got completely stuck. They just can’t see how to move forward.
By the end of our first session, they’re usually telling me how much better they feel already because we’ve stripped everything back to a point where they feel on top of things again and they have a plan.
Here’s the process we follow to do that over the time that we work together.
1 – Nail your research questions
If you’re not crystal clear on what you want to find out then you’re not going to be able to analyse your data.
You might think you already know, but so often our research questions just aren’t clear enough. I talk about this a lot in my content – and that’s because I really believe that the research question is the key to nailing your entire project!
First of all, what’s your why? What do you want to find out? What will it change?
Next, have you actually written down your research questions? It seems obvious, but a lot of my clients are working from memory. So that changes immediately when we start working together!
After that, we compare the questions with the PICO framework (population, intervention, comparator, outcome) to make sure they’re detailed enough. Often writing your research question isn’t just a one-off job. We have to go back and tweak and refine as we realise that there’s still uncertainty in there.
If you’re not working with someone on your research question, I’d recommend getting a peer review at this stage too. There’s nothing like a fresh pair of eyes to notice the little details that you’ve missed.
For more detail on how to compose the best research question for your study, check out my blog on this topic.
2 – Get your data ready for analysis
Messy data is hard to analyse and will create mistakes. That’s why my clients and I spend some time making sure it’s ready for analysis before they start.
Cleaning up the data in Excel generally involves things like:
– Making sure there’s just one piece of data per cell
– Adding identifiers to the rows
– Checking spellings
– Sense checking the data that’s been collected.
It’s important to keep a detailed audit trail of everything you’ve changed. I usually recommend using a statistical package for this reason. If you use syntax to clean your data then you’re writing your audit trail at the same time. I’ve got a blog post that takes you through the more detailed steps on how to do this.
3 – Plan your analysis
Next up, go through each of your research questions in turn and choose which test you’re going to use for that question. You’ll also need to plan how you’ll set the test up. Think about things like deciding what your outcome and explanatory variables are. These things are all very closely related so you’ll do them at the same time. I’ve got a free guide that takes you through this in more detail.
4 – Crack on with your analysis
Now you’ve got a solid plan, it’s time to put it into action. By now, you should be really clear on what you need to do because you’ve got it all planned out. You might still have some challenges ahead of you, like figuring out how to use a stats package like SPSS. But it’ll be so much easier to figure these out now you only have one part to get your head round.
(Hint: Google can help you with how to apply this in your stats package of choice!)
5 – Figure out what it all means
Your next job is to interpret the output and use it to answer your beautifully crafted research questions. There are a ridiculous number of different types of tests or models that you can apply. That makes it hard to be specific here.
Your job is to find out if the effect size is clinically important and statistically significant (I’ve spoken about this before so have a read of this). Your research paper is your place to do this interpretation, particularly in the results section.
Tailoring the programme to you
Even though these are the basic steps, I tailor the programme to address the specific needs of each client. Remember, everyone I work with will have a different starting point. It’s important to me that I take this into account when helping them. We work together at the client’s pace. Even if their project isn’t completed within the time that we work together, they’ve got a solid understanding of how to carry on.
Instead of telling you all the different possibilities of what you might do at this point, I thought I’d ask some of my most recent clients to sum up their experience.
Scott Willis, Research Associate at Loughborough University, said: “We were particularly impressed with Danielle’s ability to convey knowledge in a simple manner which was easy to understand. Additionally, Danielle was quick to respond, friendly and able to adapt her expert advice to suit our needs. Working with Danielle was extremely helpful, and I would highly recommend her services to others.”
Sundus Malaikah, Research Associate at Loughborough University, added: “While working on my research, I had to use a statistical test that was unfamiliar to me which led us to contact Danielle from Simplified Data. Now that we’ve covered all the basics about Generalized Linear Models, I feel more confident working on my research due to the help we have received from Danielle. What I liked best was how smooth and simple everything went. Statistics can look daunting to newbies but Danielle provided answers in very simple terms.”
Dr Emer Brady, Postdoctoral Fellow at University of Leicester, commented: “The consultation was perfect for my situation given I have good command of the statistical software SPSS and a good grasp of statistics but needed guidance on the approach to take and confidence in the statistical tests I was to apply in order to correctly answer the research questions in hand with the data I had. I have now embarked on the world of SPSS syntax – wow- what a time saver! In addition, Danielle provided more detail on the type of relationship I was exploring with my data with useful and trustworthy links, which gave me more confidence in what I was doing, as we all know statistics can be rather overwhelming! I then went away and conducted the analysis, which Danielle again reviewed and provided some very useful feedback – I am now just finalising this piece of work!”
The real secret of statistics
As you can see, the projects I work on are varied and different every time. It’s one of the things I love about my work. Tailoring the programme to each client is vital to providing the very best value from their investment in me.
I suppose that is the secret really. Deciding which model to use for your data is a lot like the way I tailor my programme each time. It can take a while to get right, but when you know how, the results can be phenomenal.
Are you feeling stuck with your statistics?
Maybe you’ve got a bit lost down the rabbit hole, or are using a model that’s unfamiliar. If you’re not sure how to proceed, my statistics coaching could help you move forward with confidence.
To find out if you’d be a good fit for the programme, get in touch today: email@example.com. I can’t wait to speak to you!
If you enjoyed this article, here’s a few more that you might find interesting too!