METHODS 2290F - FIELD WORK & WRITING ROTATION

if we have a dependant variable that is continuous and an independent variable that is continuous, what type of statistical test do we use?

a correlation test

If dependant variable = a CONTINUOUS variable

And

Independent variable = a CONTINUOUS variable

Then

We do a CORRELATION TEST

IF THEY'RE BOTH CON'S, THEY'RE ROTTON TO THE CORR!!!

what specific values do we report as the results of a correlation test?

Report as: (Pearson correlation coefficient (R) = __, N = __, P = __)

CORRELATION

CORRDAE!!

RNP!!

(R, N, P)

In the results section of our correlation test, what should we write?

we should write:

1. a conclusion sentence

2. sentence ending with the (Pearson correlation coefficient (R) = __, N = __, P = __) in brackets

3. brackets ending with the figure of the graph that displays the test results

if we have a dependant variable that is continuous and a categorical variable with exactly two different treatments as the independent, what type of statistical test do we use?

TWO TREATMENTS?

T TEST!!

What test statistics are calculated from a t-test?

the t

the df (degrees of freedom)

and the p

Report as: (t = __, df = __, P = __)

TEST DRIVE PORSCHE'S :)

for a t-test, if we have a low test statistic ("t") and a low degrees of freedom ("df"), should our p-value be high or low?

The p-value should be high.

- When the t-statistic is close to 0, the p-value will be large (closer to 1).

a low t-statistic: indicating a small or negligible difference

low degrees of freedom: indicating greater uncertainty due to a small sample size

the probability of that result occurring by random chance under the null hypothesis is high.

Therefore, the p-value will be high.

what does a low t-statistic indicate?

Low T-statistic (approximately 0):

- indicates that the observed difference is small relative to the variability in the data.

suggests the null hypothesis is true.

what does it mean if we have a low degrees of freedom (df)?

- Degrees of freedom are related to your sample size

- A low df means you have a low sample size (less information) and greater uncertainty

- This is reflected in the shape of the t-distribution , which has thicker tails when df is low.

so, we'd need a higher t statistic score in order to reach the same p value as another study with a higher degrees of freedom

for a t-test, if we have a high test statistic ("t") and a high degrees of freedom ("df"), should our p-value be high or low?

p-value should be low (very close to 0).

- result is statistically significant, and you reject the null hypothesis.

what does a high t-statistic indicate?

- large t-statistic: the observed difference is very far out in the tails of the t-distribution.

- the null hypothesis is extremely rare.

- results in a low p-value (reject the null hypothesis!)

what does it mean if we have a high degrees of freedom (df)?

high df = Large Sample Size

- When the df is high, the t-distribution curve becomes taller and thinner in the center and has thinner tails.

- contributes to a lower p-value

what is Standard Deviation?

measures the amount of variation in a sample

- The typical amount of variation or "spread" for individual data points around the single sample mean.

- A descriptive statistic; it does not measure precision between samples.

what is Standard Error?

standard error = (standard deviation)/(sq root of # of samples)

- it estimates how precise the estimate of the mean for the population is

- A small SE indicates that the sample mean is a reliable and precise estimate of the population mean, meaning that if you were to repeat the study, the sample means would likely be very close to each other and the true population mean.

when reporting t-test results, why is it typical to report the standard error and not the standard deviation?

- SD remains relatively constant regardless of sample size. The SE, however, incorporates sample size (n) in its calculation

ERROR BARS OF SE: typically, if they do not overlap, that means there's a significant difference between the samples

ERROR BARS OF SD: they may overlap even when the data is not significantly different

when writing ur results section, how should you report the results of a t-test?

The goal is to communicate three main pieces of information:

1. What you did (the type of test and what it compared).

2. The descriptive data (the means and variability of the groups).

3. The inferential statistics (the test values and the conclusion).

example: "Focal fish had no preference for full-siblings over unrelated individuals when the focal fish was familiar with both stimulus groups (t = 0.57, df = 22, P = 0.57; Fig. 2)"

if we have a dependant variable that is continuous and a categorical variable with more than two different treatments as the independent, what type of statistical test do we use?

MORE THAN TWO TREATMENTS?? Y'ALL GOT ANUTHA? (ANOVA)

ANOVA!!!

anova stands for analysis of variance

- Global Test: ANOVA performs one single "global" test to determine if there is a statistically significant difference among the means of all the groups simultaneously.

what test statistic results will an ANOVA test produce?

(F = ___, df = first df, second df, P = ___)

ANUTHA CRIME?

FUQ DA POLICE!!!

- A large F-ratio suggests the differences between the group means are greater than what would be expected by chance.

the df values should be:

first df = the # of groups minus one

second df = the # of samples minus the # of groups

which statistical test features two df values? what do each of these values represent

an ANOVA test presents its results with two degrees of freedom values (df = first df, second df)

first df = the # of groups minus one

second df = the # of samples minus the # of groups

if we have 4 treatments, each treatment having 3 samples, what would the df portion of our ANOVA results look like?

(df = first df, second df)

first df = the # of groups minus one

second df = the # of samples minus the # of groups

(df = 3, 8)

whats an "alpha" value?

An alpha value (α) is a value representing the probability of a false positive

– it's often set to 0.05 (5%)

- In hypothesis testing, if a p-value is less than alpha, results are considered statistically significant

What is a post-hoc test? when do we need one? how do we do them in this course?

- If ANOVA shows significance, we do post-hoc tests to find which specific groups are different. ANOVA alone won't tell us :(

- we use excel to do multiple t-tests between each possible pair of treatments

- only report the p-values, not the individual t-statistic or df for every comparison

- instead of using the typical alpha value of 0.05, we compare our p-values to a NEW, adjusted alpha

- the alpha = (0.05)/(# of t-tests done)

the # of t-tests done is how many total comparisons we're making (so, if we have 3 treatments, we'd run 3 t-tests. if we have 4 treatments, we run 6 t-tests and would have an alpha value 0.0083)

how does the alpha value change during post-hoc tests?

instead of using the typical alpha value of 0.05, we compare our p-values to a NEW, adjusted alpha

- the alpha = (0.05)/(# of t-tests done)

the # of t-tests done is how many total comparisons we're making (so, if we have 3 treatments, we'd run 3 t-tests. if we have 4 treatments, we run 6 t-tests and would have an alpha value 0.0083)

In the results section of our post-hoc test, what should we write?

- We should state which two groups have a significant difference and which group has a higher result

- State the p value in brackets after

Example: "There was a significant difference between species richness for Treatment A and for Treatment B, with Treatment A having more species (p = 0.011)".

when we're writing a results section for an ANOVA test (with significance), what should we put in the results section overall? which p-values should we include?

- we state the results from the ANOVA test and declare that there was significance among treatments

- we include the three values from the ANOVA test (in brackets, the F, df, and P = __)

- then, a new paragraph stating ALL OF THE ANNOYING POST-HOC P VALUES BTWN EACH AND EVERY GROUP

- interpret the significance between groups as a sentence and finish each sentence with the p value in brackets.

OR, sometimes for the post-hoc results we just put a table with a column of the p-values & a column of treatments we're comparing

example:

"There was a significant difference in species richness among treatments (F = 4.32, df = 3, 8, P = 0.037; Fig. 1)

Post-hoc tests revealed differences between the following treatments: Location 1 and Location 2 (p = 0.003), Location 1 and Location 3 (p = 0.001), Location 1 and Location 4 (p < 0.001), and Location 2 and Location 4 (p = 0.005). However, there was no difference in species richness between Location 2 and Location 3 (p = 0.054), or between Location 3 and Location 4 (p = 0.113)."

typically, we state a p-value with ___ decimal places

3 decimal places

ex. p = 0.113

what do we do if we wanna state a p-value that is smaller than 3 decimal places? (ex. p = 0.000278)

this is the only time when its okay to state an inequality sign (<) in our results.

we'd say: (p < 0.001)

what chart would we use to present the values of a correlation test, ANOVA test, or t-test?

Correlation Test = measures the strength and direction of a linear relationship between two continuous variables.

Chart Type: Scatter Plot

ANOVA and T-Test

- Both will have results that compare the means and variability of treatments

Bar Chart: most common

bar height is the dependent variable value

include Error Bars (usually SE of the Mean)

In a poster, everything is similar to the structure of a research paper, except there's...

no abstract section in a poster

What parts make up the structure of a research paper? how many parts? what are they each called?

- there are 9 different parts that make up a research paper

- Mnemonic: "Trevor And Adam Immediately Made R'ed Data And Ran"

Title

Authors

Abstract

Introduction

Materials and Methods (or just "Methods"

Results

Discussion

Acknowledgements

References

Explain the "abstract" section of a research paper. What does it include?

It briefly summarizes the paper

- typically includes sections from each part of the paper, briefly taking you through everything

Has sentences related to:

Introduction: we give some context and include our research question/hypothesis.

Methods: our general approach of how we ran our experiments. avoiding too much detail.

Results: state what we found! (only in words). do not include statistics (usually no references).

Discussion: state why these findings findings r important. give some context (e.g., how the research advances the field).

Explain the "introduction" section of a research paper. What does it include?

- Starts with general context for a wide audience and gets more specific as it goes on

- context leads towards the stating of the paper's objective

- Then we state the research question

- Then the hypothesis (general) and predictions (specific to the experiment).

Explain the "Materials and Methods" section of a research paper. What does it include?

- Describe the procedures so the experiment can be replicated by a reader.

- if uniquely important equipment was used, mention where you acquired it

- Only mention necessary details that impact reproducibility (e.g., incubation temp. and time).

- include the specific tests (ex. t-tests, ANOVA), you used to analyze data

- identify the dependent/independent variables of the study

- reference studies that used similar methods so u avoid repeating every detail.

- if u excluded any data, use this section to explain why (like, explain why u took an outlier sample out of ur final dataset)

- No statistics should be mentioned (put that in the results section)

if u excluded any data, use the ________ section of a research paper to explain why (like, explain why u took an outlier sample out of ur final dataset)

the "Materials and Methods" section!

We use the ________ section of a research paper to describe the procedures so the experiment can be replicated by a reader

the "Materials and Methods" section!

We use the ________ section of a research paper to state the predictions of our experiment

the introduction section!

We use the ________ section of a research paper to state a short summary of our paper

abstract!

Explain the "Results" section of a research paper. What does it include?

State what was found in the experiment!

- Results of statistical tests (ex. Plants were significantly taller (T = 2.7, DF = 39, P = 0.011; Figure 1).)

- DON'T put any biological meaning of our results (thats for the Discussion.)

- say if results are statistically significant or not (e.g., stating a result was "significantly taller").

- List ur sample sizes

- Results are presented either as a table or a figure (graph), not both.

We use the ________ section of a research paper to list our sample sizes

results!

Figure captions appear ______ the figure, while table captions appear ______ the table. (ABOVE? OR BELOW? EITHER?)

Figure captions appear below the figure, while table captions appear above the table.

Explain the "discussion" section of a research paper. What does it include?

Interpret our results, biologically!! (not statistically).

- state if results support the hypothesis and explain why.

- Relate findings to other studies (citing references) and discuss the broader meaning.

- Discuss questions for future research and how the findings could be applied.

We use the ________ section of a research paper to include the statement, "higher temp led to faster growth because there were faster moving molecules"

the discussion section!

that's a biological interpretation of our results!! we CANNOT put it in the results section (don't even think abt it.)

T or F: its okay to include statistics in the research paper's "discussion" section

False.

in the discussion, you do not restate p-values, test statistics, or sample sizes.

Statistics belong in the Results section only.

T or F: its okay to include statistics in the research paper's "results" section

True.

It's literally required to include statistics in the Results

T or F: Your Introduction should start with your study species.

FALSE

Starting with your species is bad.

- It scares away most of your audience because they think the research only applies to that organism.

- You must start broad, then gradually narrow down to your species at the end.

T or F: It's okay to include references in the introduction.

TRUE.

You should cite references in the introduction to show what is already known and what gaps remain.

T or F: You should include the statistical test name (e.g., ANOVA, t-test) in the Methods section.

TRUE.

- Include the name of the statistical test in the Methods and how it is performed.

You must write:

“We used a t-test to compare…” or

“We used an ANOVA to examine…”

But NO statistics (numbers) go here — only the test name.

What is a primary article?

Provide direct evidence about an event, object, or experiments.

- usually in the form of a journal article

- presents new data collected by the authors

- It explains how the data were collected (Methods)

- It shows what they found (Results)

- It often includes figures, tables, statistics

Ex. Experiments, Observational Studies, Mathematical Models,.

what is an identifying characteristic of primary articles?

The single most important identifying characteristic of a primary article is:

⭐ It contains a Methods section and a Results section with original data.

a primary research paper:

✅ It presents new data collected by the authors

✅ It explains how the data were collected (Methods)

✅ It shows what they found (Results)

✅ It often includes figures, tables, statistics

What is a secondary article?

Describes, discusses, analyzes, or comments on primary sources.

- Rarely include methods and results sections.

- (ex. Reviews, Books, Scientific Magazines.)

what's a meta-analysis?

taking a bunch of studies and trying to get a final conclusive result

- complex type of secondary study that combines data from several published studies to find an overall pattern.

- Often have a Method section (describing how articles were found) and a Results section (reporting standardized effect sizes).

- almost always report effect sizes, not raw means.

T or F: a meta analysis is a primary article

FALSE

- even tho they often have methods/results sections, they're NOT primary

- it combines data from several published studies to find an overall pattern, but this publisher did not collect any data themselves

- They are secondary because the authors did not collect the data themselves.

list the three types of primary and two types of secondary journal articles

Primary articles:

- experiment

- observation

- model

Secondary articles:

- meta-analysis

- review

what is an experiment?

Manipulates one or more variables (treatments) to determine the effect on a response variable.

- either done in a lab or in nature (field studies)

Field experiment (still an experiment):

Researchers assign different fertilizers to plots in a real farm

Crop yield is measured

Happens in nature ✔

Treatment is manipulated ✔

Full Factorial Design: Necessary in any experiment when manipulating multiple variables; all combinations of treatments must be tested to "tease apart" the individual and combined effects of each factor.

in any experiment when manipulating multiple variables, what must we do?

we must incorporate a full-factorial design! this means we test every possible combination of the independent variables

- all combinations of treatments must be tested to "tease apart" the individual and combined effects of each factor.

what's the point of making our experiment a "Full factorial design"?

“Without a full factorial design, you don’t know whether the observed difference is from variable A, variable B, or the combination of the two.”

- When multiple variables are manipulated (e.g., temperature + nitrogen), every combination must be tested:

Low T / Low N

High T / Low N

Low T / High N

High T / High N

➡️ clarifies independent vs combined effects.

what's an Observational Study?

- a primary research study that uses natural difference/variation in conditions to look at environmental effects on subjects. (e.g., effect of polluted vs clean rivers on guppy health).

- The environment is often chosen to be similar in most ways except for one important variable of interest (e.g., comparing guppy coloration in two different rivers).

- No manipulation we do on variables

- Must acknowledge confounding variables.

Important lesson: ➡️ Observational studies can suggest patterns but cannot prove causation.

what is a "model"?

A model (often a Mathematical model) is simplified version of reality built with mathematical equations to test understanding and make predictions

✔ Uses equations

✔ Represents reality in a simplified way

✔ Allows predictions without physically running the experiment

✔ Often used when experiments are impractical, slow, or impossible

- representation of an object or event

- Simplified simulations of real systems using equations.

(ex. a model shows that, if CO2 emissions stay steady until 2050, sea levels will rise 10m)

Purpose: ➡️ Test how systems behave under different assumptions.

What is an "effect size" and where do we see them?

an effect size is what a meta-analysis will report

effect size measures the difference between the effect of treatments

The larger the effect size, the bigger the difference between treatments.

if we have a mathematical model of global warming and we publish a paper stating "earth is predicted to raise 4 degrees in the next decade", is this a primary or secondary research article?

a primary article.

- we generated new results — even though the “data” are simulated, not measured.

- Even though they don’t collect field/lab data, mathematical models create new predictions

The larger the effect size, the bigger the...

The larger the effect size, the bigger the difference between treatments.

- a meta analysis will report this

what is a review paper?

a summary of what is known about a subject

- this is a SECONDARY article

- Summaries of what is known.

- Helps scientists identify gaps in the field and impacts future research directions

How does the scientific method occur? explain all components

Our restaurant has pizza-eating races regularly

1. begins with an observation

- this may be something that interests us or something strange we noticed

2. then, we make a research question

- You turn the observation into a question that can be studied.

- Why did that strange observation occur?

3. we develop an educated guess based on prior knowledge, and formulate a hypothesis

- we create a possible explanation that can be proven or falsified with a designed study (test)

4. We design our experiment but don't do the experiment until we have a prediction on what will occur in the study

- Predictions describe what you expect to see if the hypothesis is true.

- They are specific and measurable.

5. we conduct the test/experiment to see if our prediction is right

6. we then report our results

7. and move on to do any recommended action

After interpreting results, you explain: why they matter, what the implications are, what should be done next (applications, future studies, decisions).

if we have the hypothesis, "Higher temperatures increase plant growth." then what might our "prediction" be?

remember that, in the scientific method, the prediction step comes after the hypothesis

Predictions describe what you expect to see if the hypothesis is true.

They are specific and measurable.

Example:

If plants are grown at 25°C vs 15°C, the 25°C plants will be taller after 10 days.

Notice:

➡️ The hypothesis is general.

➡️ The prediction is concrete.

according to the scientific method, what do we do if the results of our tests do not support the hypothesis?

If the results do not support the hypothesis →

➡️ You formulate a new hypothesis with new predictions and test again.

This is GOOD science — hypotheses are not failures when rejected.

after a scientist does their experiment, writes a conclusion, and makes a research article, what do they do next? (explain the whole process that occurs after you write an article)

Peer Review and then Publication

- A scientist writes a paper, submits it to a journal editor, who forwards it to typically two or three peer reviewers.

- Reviewers Can be professors, government/industry scientists, or even graduate/undergraduate students, often selected for their expertise and lack of conflict of interest.

- The editor issues a decision based on the reviewers' recommendations (e.g., accept, reject, reject with major/minor revisions).

- Being rejected does not mean the paper is bad; it may be a bad fit for the journal's audience or impact level. Reviewer comments are crucial for improving the paper for resubmission elsewhere.

What Are Blogs & News Stories in Science?

- They are NOT peer-reviewed scientific sources.

- They are NOT primary or secondary articles.

- They are a form of public science communication.

Their purpose is to:

- spread new scientific findings quickly (getting peer-reviews takes forever, posting about your research on twitter can be done in seconds)

- make research understandable to a general audience

- stimulate discussion

- They translate science, not perform it.

Which test creates an "F" test statistic? what does it represent and how can we interpret its value?

The F test statistic is produced by ANOVA (Analysis of Variance)

a large F value indicates that the group means differ more than would be expected from random variation alone.

- big F = tells us that at least one population mean is different, but it does not tell us which one.